Agentic AI
Terminal Coding Agent
A coding agent for the terminal that reads, edits and runs code, with plan mode, MCP tools, subagents and long-term memory, on any OpenAI-compatible model.
Senior AI Engineer · Chubb
Abinav Ramesh
I design and ship production Gen AI systems: RAG pipelines, enterprise LLM assistants, agentic workflows and AI governance platforms, built on 7+ years of machine learning and analytics.
- $5MM
- value delivered by an Agentic Enterprise Assistant at TD
- $10M CAD
- saved each year by liquidity forecasting models
Projects
Gen AI & agents
Recent work on LLM applications, multi-modal models and agent systems.
LLM Applications
Conversational Text2SQL
Ask a database a question in plain English and get ready-to-run SQL and its results, with an LLM-as-judge evaluator to measure accuracy.
Agentic AI
Agentic AI Financial Agent
A team of specialist agents that pull live market data, search and scrape the web, and summarise videos into one sourced answer.
Multi-Modal LLMs
DocTalk: Multi-Modal Chat App
One chat for PDFs, code, images and audio, with Gemini reasoning over the raw file instead of a chain of OCR and transcription steps.
LLM Applications
Smart Resume ATS System
Upload a resume and a job description to get a structured candidate summary and a skill-gap analysis in seconds.
LLM Applications
LLM Chatbot
A from-scratch multi-turn chatbot on Gemini 2.0 Flash with conversation memory and a model layer that can be swapped out.
Multi-Modal LLMs
Nutritionist LLM
Upload photos of everything you ate in a day and get the food items, calories, macros, a 1-10 health score and suggested swaps from a vision model.
Earlier work
Machine learning, NLP and data engineering projects from before the LLM era.
Text Mining
Car Brand Analysis
Scraped Edmunds forum posts and used lift ratios and an MDS map to show which car brands buyers see as similar, and what they associate with each.
Data Engineering
Data-Triggered Workflow in Apache Oozie
An Oozie coordinator that starts the daily workflow only after three days of one feed and two days of another have fully landed.
Developer Tooling
Client-Side Git Hooks
Client-side hooks that auto-format Python with Black, enforce commit-message rules and block unwanted rebases, installed with one script.
NLP
Legalisation of Marijuana in Canada
Province-by-province Twitter sentiment before and after cannabis legalisation, weighted by follower reach, to recommend how to roll out controversial policies.
NLP
Increasing Engagement on National Geographic's Instagram
Predicted Instagram engagement from Google Vision image labels and captions (77% accuracy), then used LDA topics to show that wildlife posts drive the most engagement.
Automation
SFTP Transfer Automation
A scheduled Python job that pulls data from Impala, writes a dated CSV and delivers it to the right folder on an SFTP server, with no manual steps.
NLP
Salary Prediction from Job Descriptions
Compared Naive Bayes models on job-posting text, job attributes and both combined to predict high versus low salary. The text-only model won with 77.2% accuracy.
Social Network Analysis
Network Analysis of the AirPower Cancellation
Analysed about 30,000 tweets as Apple cancelled AirPower: sentiment, emoji use, how the network exploded, and which influencers spread it worldwide.
Computer Vision
Face Detection on FPGA
Real-time face detection prototyped in MATLAB and deployed to a ZedBoard with a Xilinx Zynq-7000 SoC.
About
Experience
I build Gen AI systems that hold up in production: agents, agentic workflows and the RAG pipelines underneath them. Seven years of machine learning and analytics taught me to judge every system by the business impact it delivers.
-
Nov 2025 – Present
Chubb
Senior AI Engineer
Building AI Studio, an enterprise platform for AI governance of MCP tools, agents and models - designing APIs and microservices, integrating Dify and Azure services for secure, governable GenAI use, and embedding governance guardrails (access, lineage, monitoring) into the agent/model lifecycle.
-
Aug 2021 – Dec 2025
TD Bank Group
Data Scientist → ML Engineer (Layer 6 AI) → Senior Applied ML Scientist
4.5 years in Personal Banking Data & Analytics. Delivered $5MM in value building an Agentic Enterprise Assistant on Azure OpenAI and Azure AI Search, grounded by a RAG pipeline (query rewriting, ranking, insufficiency detection). Built a standardized RAG framework that cut application build time from 12 to 7 weeks, and time-series liquidity forecasting models saving $10M CAD annually.
-
Oct 2020 – Sep 2022
The Bigger Data
Data Science Project Manager
Led a team of 5 at The Bigger Data to deliver analytics solutions for non-profits - improving operational efficiency by 30%, data accuracy by 90%, and building governance frameworks that avoided $200K in potential regulatory fines. Worked with AIESEC to improve their business model through analytics.
-
Jul 2019 – Aug 2021
Bell Mobility
Data Science Developer
Built critical AI/ML products: a modem self-healing system that saved $1.1M CAD by proactively detecting and fixing degraded customer experience, proactive network outage prediction, automated reboot workflows resolving 40% of modem issues, and MicroStrategy dashboards driving $500K in annual savings.
-
Aug 2018 – Jul 2019
McGill University
Master of Management in Analytics (MMA)
GPA 3.92, Dean’s Honour List, Best Capstone Project and Best Outgoing Student Award. Capstone: Data Engineer/Modeller intern at McKesson Canada, building a Supplier Performance Management Program that saved up to $600K.
-
Aug 2016 – Jul 2018
Deloitte Consulting
Business Technology Analyst
Worked across the full application lifecycle for US clients: converted 50+ OBIEE reports to SAP Business Objects on a SAP HANA backend, resolved 500+ production tickets as a Production Support Engineer with 100% SLA compliance, and served as a System Integration Tester for a major healthcare insurance provider. Received the Applause Award for outstanding performance.
-
Jul 2012 – Jul 2016
National Institute of Technology Karnataka
B.Tech, Electronics and Communication Engineering
Toolbox
Skills
Generative AI
Deep Learning / NLP
Cloud
Big Data / DevOps
Data Science / Visualization
Databases
Recognition
Certifications
Ericsson Data Science Hackathon
First place, Montreal
AWS Certified Cloud Practitioner
Amazon Web Services
Microsoft Azure Fundamentals
Microsoft
McGill MMA Ambassador
Desautels Faculty of Management
Courses
- Kubernetes for BeginnersUdemy
- SQL Server Machine Learning Services: PythonLinkedIn Learning
- Big Data Modeling and Management SystemsCoursera
- Data Visualization and Communication with TableauCoursera
- Managing Big Data with MySQLCoursera
- Business Metrics for Data-Driven CompaniesCoursera
- Predictive Customer AnalyticsLynda.com
- Agile Development PracticesLynda.com
Contact
Let's talk Gen AI
I'm open to Gen AI, machine learning and data roles and collaborations. The quickest way to reach me is email or LinkedIn, or you can book a 15-minute call.