<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Maria Philna Aruja on Medium]]></title>
        <description><![CDATA[Stories by Maria Philna Aruja on Medium]]></description>
        <link>https://medium.com/@maria.aruja?source=rss-3a43092ee68a------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*sNoJCx3ouWWUVUkY</url>
            <title>Stories by Maria Philna Aruja on Medium</title>
            <link>https://medium.com/@maria.aruja?source=rss-3a43092ee68a------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 09:31:30 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@maria.aruja/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Enterprises Don’t Have a Data Problem — They Have an Access Problem]]></title>
            <link>https://medium.com/quantrium-tech/enterprises-dont-have-a-data-problem-they-have-an-access-problem-bfe9ee22bcca?source=rss-3a43092ee68a------2</link>
            <guid isPermaLink="false">https://medium.com/p/bfe9ee22bcca</guid>
            <category><![CDATA[enterprise-technology]]></category>
            <category><![CDATA[ai-chatbot]]></category>
            <category><![CDATA[secure-access]]></category>
            <category><![CDATA[mcp-protocol]]></category>
            <category><![CDATA[agentic-rag]]></category>
            <dc:creator><![CDATA[Maria Philna Aruja]]></dc:creator>
            <pubDate>Sun, 01 Feb 2026 05:29:47 GMT</pubDate>
            <atom:updated>2026-02-01T05:29:47.456Z</atom:updated>
            <content:encoded><![CDATA[<h4>QUANTRIUM GUIDES</h4><h3>Enterprises Don’t Have a Data Problem — They Have an Access Problem</h3><h4>Why Secure, Agentic Access Is the New Enterprise Interface</h4><figure><img alt="A background image with hands on a laptop overlayed with Secure Trust Icon with Icons and text for Secure Agentic Access, Retrieval Augmented Generation, Model Context Protocol and Continuous Feedback." src="https://cdn-images-1.medium.com/max/1024/1*1GfA9MQ9hIbfHbNmejhlNg.png" /><figcaption>Secure Agentic Access — New Enterprise Interface</figcaption></figure><h3>The Missing Link: Accessing Enterprise Knowledge</h3><p>Not lack of information, but access to information, is the biggest productivity problem in modern enterprises.</p><p>Here’s a scenario that plays out in offices everywhere.</p><ul><li>A project manager is looking for information on the technology stack for a client’s project.</li><li>A new hire wants to understand the company’s leave policy.</li><li>A sales lead is searching for last quarter’s performance metrics.</li></ul><p>The information exists. Somewhere. Buried in a PDF sitting in someone’s Google Drive. Locked away in a PostgreSQL database that only the IT team can query, or fragmented across Confluence pages, SharePoint folders, and legacy systems that haven’t been updated in months.</p><p>Every enterprise sits on a goldmine of internal data, such as HR policies, project documentation, technical specifications, and financial reports. But that data lives in silos: PDFs, DOCX files, HTML pages, relational databases like MySQL and SQL Server, and NoSQL systems like MongoDB.</p><p>When someone needs an answer, they usually face two poor options.</p><p><strong>Option one:</strong> Spend hours hunting through folders and documents, hoping the version is current.</p><p><strong>Option two:</strong> Raise a ticket with IT and wait for someone to write and validate a SQL query.</p><blockquote>Neither of these scales.</blockquote><p>For non-technical teams, this friction is constant. Why should a marketing manager understand database schemas? Why should an HR specialist navigate complex folder hierarchies to answer routine questions?</p><p>This isn’t just inefficiency; it’s a fundamental disconnect between the data organizations already have and their ability to actually use it.</p><blockquote>This is where the Enterprise Agent comes in.</blockquote><h3>From Search to Conversation and Then to Action</h3><p>Traditional enterprise search tools stop at retrieval. They help you <em>find</em> documents, but do not <em>understand</em> or <em>use</em> the information inside them.</p><p>Large Language Models changed expectations. Employees now expect to ask questions in natural language and get direct, contextual answers. But deploying LLMs naively inside an enterprise quickly exposes real limitations:</p><ul><li>Hallucinated or incomplete answers</li><li>Outdated context</li><li>No grounding in internal systems</li><li>Serious security and compliance risks</li></ul><p>Retrieval-Augmented Generation (RAG) addresses part of this problem by grounding responses in enterprise data. But <strong>RAG alone is not enough</strong>. Enterprises need orchestration.</p><ul><li>They need governance.</li><li>They need agents that can reason across systems — not just retrieve text.</li></ul><h3>How Quantrium Is Solving the Access Problem</h3><p>At Quantrium, we’re addressing this challenge in a phased, pragmatic way, starting where enterprises feel the pain most and scaling systematically.</p><p>At the core of our approach is a simple idea: enterprise knowledge should be unified semantically, not physically.</p><p>The Enterprise Agent platform is built around three foundational layers:</p><ol><li>Data Ingestion and Integration</li><li>Retrieval and Reasoning with RAG</li><li>Continuous Feedback and Improvement</li></ol><h3>Structured Data Without the Bottlenecks</h3><p>Phase One delivers production-grade access to structured data stored in SQL databases, specifically MySQL, PostgreSQL, and Microsoft SQL Server. This is where critical business data often lives and where access is typically restricted to a small technical group.</p><p>Our goal is straightforward:</p><blockquote>“Make structured data accessible to non-technical users without compromising security, control, or accountability.”</blockquote><p>The Enterprise Agent platform securely connects to existing enterprise data sources and creates a unified semantic representation of enterprise knowledge. Raw data never leaves the enterprise boundary. Only schemas, metadata, and contextual signals are embedded for retrieval.</p><p>Phase One emphasizes secure access, schema-aware query generation, full observability, and simple containerized deployment within the enterprise’s environment.</p><h3>How It Works Under the Hood</h3><p>The system uses a Retrieval-Augmented Generation (RAG) pipeline designed for schema-aware reasoning. During ingestion:</p><ul><li>Database schemas (tables, columns, relationships, metadata) are read</li><li>This information is chunked and embedded</li><li>Embeddings are stored in Qdrant</li></ul><p>No raw data is exposed externally; only contextual representations required for retrieval are accessed.</p><p>When a user submits a natural-language question:</p><ol><li>The query is embedded</li><li>Relevant schema context is retrieved</li><li>The LLM generates a SQL query grounded in enterprise-specific schemas and constraints</li></ol><p>Users can either:</p><ul><li>View and download results directly through the application, or</li><li>Receive the generated SQL query itself, for teams that prefer manual execution</li></ul><h3>A note on accuracy</h3><blockquote>Like any AI-assisted system, generated queries require review. Complex joins or nuanced business logic may need iteration. The interface is intentionally designed to support verification rather than blind execution.</blockquote><blockquote>Either way, the dependency on IT tickets is dramatically reduced, without sacrificing oversight.</blockquote><h3>Continuous Feedback and Improvement</h3><p>The Enterprise Agent maintains conversational context and supports user feedback on generated results.</p><p>This feedback loop helps to:</p><ul><li>Improve query disambiguation</li><li>Increase first-attempt answer accuracy</li><li>Refine retrieval quality over time</li></ul><blockquote>The system is designed to improve through real enterprise usage rather than static prompts.</blockquote><h3>Expanding the Scope: From Data Access to Agentic Workflows</h3><p>From structured data, the system expands incrementally:</p><ul><li><strong>Phase Two:</strong> Question answering from PDFs and unstructured documents</li><li><strong>Phase Three:</strong> Native support for NoSQL systems such as MongoDB</li><li><strong>Phase Four:</strong> A fully agentic approach where the Enterprise Agent reasons across multiple systems and MCP servers</li></ul><blockquote>And this is where the real enterprise challenge emerges — because answering real questions rarely involves a single system.</blockquote><h3>The Integration Problem: Why Connectivity Matters</h3><p>Enterprise knowledge doesn’t live in one place. A question like <em>“What’s the status of the Johnson project?”</em> might require accessing:</p><ul><li>Project management tools</li><li>Recent documents</li><li>Repository / Release notes</li><li>Internal communications</li></ul><p>Traditionally, connecting these systems is expensive and brittle. Each integration requires custom development, ongoing maintenance, and careful handling of authentication and permissions.</p><p>As the number of systems grows, integration complexity explodes into an N×M problem.</p><blockquote>This is where the <strong>Model Context Protocol (MCP)</strong> changes the equation.</blockquote><h3>MCP: A Standard for AI to System Connectivity</h3><p>The Model Context Protocol (MCP) is an open standard introduced in late 2024 that provides a consistent way for AI systems to interact with external tools and data sources.</p><p>Instead of building bespoke integrations for every combination of AI application and system, MCP allows integrations to be built once and reused across compatible clients.</p><p>At a high level:</p><ul><li><strong>MCP Servers</strong> expose data and functionality from external systems</li><li><strong>MCP Clients</strong> (like the Enterprise Agent) consume those capabilities via a standard protocol</li></ul><blockquote>The protocol has seen rapid adoption and is now managed by the Linux Foundation’s <a href="https://aaif.io/">Agentic AI Foundation</a>, which is a strong signal of its trajectory as shared industry infrastructure.</blockquote><h3>Quantrium’s MCP Servers: Enterprise-Ready by Design</h3><p>Quantrium has developed a suite of enterprise-grade MCP servers that integrate seamlessly with the Enterprise Agent.</p><p>Each MCP server:</p><ul><li>Handles authentication and permissions</li><li>Translates queries into system-specific operations</li><li>Enforces access controls and auditing</li><li>Exposes a consistent, governed interface to the agent</li></ul><p>When an employee asks a question, the Enterprise Agent can route requests to one or more MCP servers, synthesize the retrieved information, and present a coherent response without the user needing to know which systems were consulted.</p><h3>Deployment Model</h3><p>All MCP servers are designed to run inside the enterprise’s own infrastructure:</p><ul><li>Sensitive data stays within organizational boundaries</li><li>Identity integrates with existing access management</li><li>Audit logs provide visibility into data usage</li></ul><blockquote>This is not a SaaS black box. It’s a deployable infrastructure that enterprises control.</blockquote><h3>What This Changes for Enterprise Teams</h3><p>The combination of the Enterprise Agent and MCP servers creates a practical “single interface” for everyday enterprise questions.</p><p><strong>What it enables:</strong></p><ul><li>Natural-language access to multiple internal systems</li><li>Faster answers to routine questions</li><li>Reduced context switching and tool hopping</li></ul><p><strong>What it doesn’t replace:</strong></p><ul><li>Human judgment</li><li>Review for high-stakes decisions</li></ul><p>In early deployments, the value is most visible in the long tail of routine requests: policy lookups, status checks, historical data retrieval, and cross-referencing information across systems.</p><blockquote>This is where productivity gains compound quietly but meaningfully.</blockquote><h3>Where We Are Today</h3><p>Quantrium is validating this approach through pilot deployments with enterprise clients.</p><p>The roadmap continues:</p><ul><li>Structured data access is live and evolving</li><li>Unstructured document support is being developed</li><li>NoSQL connectivity is planned</li><li>MCP server coverage is expanding across enterprise systems</li></ul><p>Each phase builds on the same core idea: enterprise data should be secure and usable.</p><blockquote>When access becomes intuitive and controlled, information stops being a bottleneck and starts becoming an accelerator.</blockquote><h3>Technical Notes</h3><p>For readers interested in implementation details:</p><ul><li><strong>LLM:</strong> OpenAI GPT-4o</li><li><strong>Vector Database:</strong> Qdrant</li><li><strong>Orchestration:</strong> Dapr Agents</li><li><strong>Backend:</strong> Python, FastAPI</li><li><strong>Frontend:</strong> Vue.js</li><li><strong>Observability:</strong> OpenTelemetry with SigNoz</li></ul><p>The MCP servers are built using the official Python SDK and follow the protocol specification for tools, resources, and prompts to ensure long-term ecosystem compatibility.</p><p>Stay tuned for more…</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bfe9ee22bcca" width="1" height="1" alt=""><hr><p><a href="https://medium.com/quantrium-tech/enterprises-dont-have-a-data-problem-they-have-an-access-problem-bfe9ee22bcca">Enterprises Don’t Have a Data Problem — They Have an Access Problem</a> was originally published in <a href="https://medium.com/quantrium-tech">Quantrium.ai</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Top 3 Packages for Named Entity Recognition]]></title>
            <link>https://medium.com/quantrium-tech/top-3-packages-for-named-entity-recognition-e9e14f6f0a2a?source=rss-3a43092ee68a------2</link>
            <guid isPermaLink="false">https://medium.com/p/e9e14f6f0a2a</guid>
            <category><![CDATA[nltk]]></category>
            <category><![CDATA[named-entity-recognition]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[spacy]]></category>
            <category><![CDATA[flair]]></category>
            <dc:creator><![CDATA[Maria Philna Aruja]]></dc:creator>
            <pubDate>Sat, 12 Feb 2022 08:15:08 GMT</pubDate>
            <atom:updated>2022-02-12T08:15:08.759Z</atom:updated>
            <content:encoded><![CDATA[<h4>QUANTRIUM GUIDES</h4><p>Comparing SpaCy, NLTK and Flair — the top 3 NER models</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/1*vXVBBFpiyIYUWbCmKSf63g.png" /></figure><p>By the time you finish reading this article, the amount of text on the internet will grow by over 2 million tweets, 15 papers in scientific journals, 4 Wikipedia articles, and countless news stories and blog posts.</p><p>To argue for the utility of such a seemingly infinite repository of text, the proponents of big-data will be quick to quip that <em>“data is the new oil”</em>, but even the richest of oil reserves require an array of tools that enable the extraction and purification of the oil to be of any use.</p><p>Named entity recognition is one such tool in the arsenal of natural language processing that allows us to tame large and unstructured text corpora. In this article, we elaborate on the inner workings of NER and discuss three widely used packages — NLTK, SpaCy, and Flair.</p><h3>Named Entity Recognition</h3><p>Named Entity Recognition is a two-step process that helps in extracting useful insights from data. It involves identifying keywords, i.e. the named entities, and then categorising them into broader classes.</p><p>For instance, in the sentence “The world’s biggest chocolate cake will be made for this Christmas by the Royal cake house.”, Chocolate cake, Christmas, and Royal cake house are the named entities in the sentence. Classifying them as say, “Food”, “Occasion”, “Organisation” is the next step in NER.</p><h3>The Process</h3><p>The process involves two stages.</p><ol><li><strong>Extraction of entities </strong>—In this step, the NER model scans the data and finds the words that can be treated as an entity. IOB tagging, Inside Outside Beginning tagging, can be used to mark the entities in the data. An NER model will be able to find the named entity in the model based on the named entities known to it.</li><li><strong>Classification of entities</strong>—This is the stage when the entities are categorised into predefined classes, the named entity “chocolate cake” would be categorised as food for instance. The effectiveness of the classification of the entities will depend upon the relevance of the training data. The more similar the training data with the type of the testing data, the more effective the NER model will be.</li></ol><h3>Packages for NER</h3><ol><li><strong>SpaCy</strong></li></ol><p>Spacy is a powerful Natural Language processing tool used to process large amounts of data. With support for over 64 languages and 63 trained pipelines for 19 languages, it is a handy tool in NLP.</p><p>It uses Bloom embedding and residual CNN’s to identify the named entities. Here is an example of NER performed using SpaCy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/791/1*wuVnGrEvt1pQjSqnW4hVsw.png" /></figure><p><strong>Output</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/429/0*eJoeq-pGgRYFHCia" /></figure><p><strong>2. NLTK</strong></p><p>Natural language Toolkit is a set of libraries used for NLP. It is widely used in research and for educational purposes. Written in Python, it has access to more than 50 text corpora across 7 languages.</p><p>One of the primary principles of NLTK, besides simplicity, consistency, and extensibility, is modularity. Modularity provides components that can be used independently. This might be especially useful for tuning only specific parts of the pipeline or even using third parties in conjunction with this toolkit.</p><p>NLTK does NER in two steps. The first step is POS (parts-of-speech) tagging or grammatical tagging, which is followed by chunking to extract the named entities. An example of NER performed in NLTK is given below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/729/0*BAQnuMe11QdM90T0" /></figure><p><strong>Output</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/334/1*5OjZnKK46omPI5iTozSzvQ.png" /></figure><p><strong>3. Flair</strong></p><p>Flair is a simple framework developed for NLP. Flair is built on top of PyTorch which is a powerful deep learning framework. Claimed to support over 250 languages, it is very useful in training small models.</p><p>It is pre-trained on an extremely large unlabelled text corpora. An example of NER performed in Flair is as given below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/729/0*bwljg0s26q8TtxU0" /></figure><p><strong>Output</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/589/0*yn5V0El_owaslQpj" /></figure><h3>Factors of Consideration</h3><p>The best package for an application can be selected based on several important factors — the availability of pre-trained model tags, the format of training data, ease of use, speed of execution, evaluation metrics, and visualisation capabilities of the package.</p><h4>Pre-trained model tags</h4><p>Pre-trained model tags indicate the capability and depth of a package. The more the pre-trained model tags, the more insightful the results will be. The available pre-trained NLTK model is only limited to 3 main tags which are PERSON, ORGANISATION, and GPE, while the available SpaCy and Flair models can identify up to <a href="https://huggingface.co/flair/ner-english-ontonotes-large">18 tags</a> using the pre-trained model.</p><h4>Format of training data</h4><p>The three packages have different formats in which they accept the training data.</p><p><em>NLTK</em></p><p>The NLTK model for custom-named entity recognition can be developed with the help of the Stanford NER tagger, written in Java. The training data for NLTK looks like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/423/1*wtqhrlo4yVTeCfiwZvGUSQ.png" /></figure><p>Each document in the training data should be separated by a newline.</p><p><em>SpaCy</em></p><p>The training data for SpaCy version 2 is a list of tuples. Each tuple contains a text, start index, end index, and a label. For version 3, refer <a href="https://towardsdatascience.com/using-spacy-3-0-to-build-a-custom-ner-model-c9256bea098">here</a>.</p><p>For example</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/525/1*TTDa4xnqYrK7Pc5dbOtjoQ.png" /></figure><p><em>Flair</em></p><p>To train an NER model using Flair, the training data should be of the following format. Each sentence in the training data should be separated using newlines. Each row has two columns. The first column is the word and the second column is the BIO-annotated NER tag.</p><p>For example</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/602/1*FLZ4-8Pr2OjhRNQre1T62A.png" /></figure><p>The performance of trained models is better for SpaCy and Flair when compared to that of NLTK.</p><h4>Ease of use</h4><p>If one package allows you to do NER in one step, you may need to execute multiple steps to perform NER in other packages. In NLTK, POS tagging needs to be done as a first step. It is then followed by chunking. This will identify the entities in the document. In SpaCy the “ents” property of the doc object identifies the named entities whereas, in Flair, the predict function is used for the same.</p><h4>Speed</h4><p>Speed is very important when you’re dealing with large datasets but there is a trade-off between speed and accuracy. The speed of a package is measured using WPS, words per second. Of the three packages, NLTK is found to be the slowest while Spacy beats Flair. So if the processing involves a very large data set, then Spacy is the best option.</p><h4>Visualisation</h4><p>Visualising the entities is an effective way of understanding the data for which the three packages have their own methods. NER entities can be visualised in NLTK using the <a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.draw.html">Draw function</a> in the Matplotlib package. SpaCy uses an inbuilt dependency visualiser called <a href="https://explosion.ai/demos/displacy#:~:text=spaCy%20also%20comes%20with%20a,directly%20from%20a%20Jupyter%20Notebook.">Displacy</a>. It can be used in Jupyter Notebook and can even be used in a browser. Flair makes use of the package <a href="https://github.com/lunayach/visNER">Visner</a> for performing visualisation. Displacy is the best option out of the three in terms of visualisation.</p><h4>Evaluation metric</h4><p>A model can be evaluated using precision, recall and F1 scores. Both NLTK and Flair have built-in functions to evaluate the model whereas SpaCy uses a scorer package for evaluation.</p><p>The table below shows the precision, recall, F1 score, and accuracy score using the three packages for a CoNLL 2003 dataset:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/674/1*ppXgFgzaFZCfZhqDlWGE_w.png" /></figure><h4>Selection of package</h4><p>By comparing the performance matrix it is observed that Flair tops the list compared to the other two. In terms of speed, SpaCy beats the other two.</p><p>Each package offers unique advantages. While it is clear that Flair and SpaCy perform better than NLTK, a choice between SpaCy and Flair can be taken considering the nature of the dataset. For smaller datasets, Flair is a good option considering the accuracy. The time taken will be almost the same as that of SpaCy. SpaCy compromises its precision for its speed.</p><h4>Limitations</h4><p>There are limitations for all three packages. It is important to know these limitations for selecting the ideal package for the application. NLTK is slower when compared to the other two packages. It doesn’t employ neural networks and splits the sentences without considering the semantics of the sentences.</p><p>SpaCy is huge, however, the package is not customisable and the internals are not opaque.</p><p>Flair is more ideal for smaller applications but its lower speed makes SpaCy a better candidate for large datasets.</p><p><strong>References</strong></p><ol><li><a href="https://monkeylearn.com/blog/named-entity-recognition/">Named-entity-recognition</a></li><li><a href="https://spacy.io/">SpaCy</a></li><li><a href="https://www.nltk.org/">NLTK</a></li><li><a href="https://towardsdatascience.com/a-pathbreaking-evaluation-technique-for-named-entity-recognition-ner-93da4406930c">Evaluating an NER model</a></li><li><a href="https://medium.com/@subshakya591/custom-trained-nltk-stanford-ner-tagger-and-spacy-ner-tagger-comparison-d4e20e41b0bf#:~:text=The%20two%20very%20popular%20tools,useful%20for%20our%20specific%20task">Training custom NER model using NLTK and SpaCy</a></li><li><a href="https://medium.com/thecyphy/training-custom-ner-model-using-flair-df1f9ea9c762">Training custom NER model using Flair</a></li><li><a href="https://realpython.com/nltk-nlp-python/">NLTK for NLP</a></li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e9e14f6f0a2a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/quantrium-tech/top-3-packages-for-named-entity-recognition-e9e14f6f0a2a">Top 3 Packages for Named Entity Recognition</a> was originally published in <a href="https://medium.com/quantrium-tech">Quantrium.ai</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>