Carbon MCP
A Python tool that checks code for potentially carbon-inefficient patterns and returns directional energy and CO₂ estimates through a CLI, API, or MCP server.
GenAI Engineer
I work on applied AI systems and build small tools around problems I find interesting.
Currently at Deloitte and based in Bengaluru. My recent work spans agent workflows, data products, developer tools, and personal software.
Selected work
A few projects that represent the kind of work I enjoy.
MGnify visualisations
My Google Summer of Code 2023 work with EMBL-EBI, exploring reusable visualisations for metagenomic quality, taxonomy, and functional analysis.
Kindle AI Highlights Organizer
A local utility that separates Kindle clippings by book, combines notes with highlights, and can summarize them using Ollama.
Recent activity
A couple of recent posts from LinkedIn. Newer activity is on my profile.
Experience
A closer look at the systems I have built, the problems behind them, and the teams I worked with.
GenAI Engineer
Deloitte · Bengaluru
I work across AI product engineering, evaluation, and enablement for enterprise teams. Most of the work sits between prototype and production: helping models use the right context, show their evidence, and fit into real workflows.
- healthcare professionals in scope
- 1,000+
- images evaluated
- 2,000+
- validated Text-to-SQL accuracy
- ~94%
- people trained across four geographies
- 400
Context and reasoning
Built graph- and agent-based systems for sales intelligence, investor research, and finance analytics. The work included multilingual inputs, entity linking, Graph RAG, source traceability, reflection, and schema-aware SQL generation.
LangGraph · Neo4j · Graph RAG · Text-to-SQL
Multimodal systems
Developed an AI-assisted embryo-grading workflow across 2,000+ images using visual retrieval, segmentation, and multimodal LLMs, with grounded evidence and readable rationales. I also built approval-aware content pipelines spanning images, narration, and video.
DINOv2 · FAISS · Multimodal LLMs · Vertex AI
Enablement and automation
Designed hands-on agentic AI training for 400 people across four geographies. I also automated commodity-risk and finance-reporting workflows, improving efficiency by more than 50% while keeping recurring delivery on schedule.
Agent frameworks · Python · SAP · Power BI
May — Aug 2023
Google Summer of Code
EMBL-EBI · MGnify
Built 16 reusable visualisations for metagenomics quality, taxonomy, and functional analysis. The Vega-Lite specifications worked across React applications as well as Python and Jupyter workflows.
React · TypeScript · Vega-Lite · Python · Jupyter
Read the final reportMar — Jun 2023
Project Intern
ITC Infotech
Worked on a climate-sensitive credit-risk model that combined financial risk analysis with climate parameters and regulatory factors using data analysis and machine learning.
Python · Machine learning · Risk analytics
Apr — Aug 2022
Project Intern
Mobil80 Solutions and Services
Worked on NLP features for SKILTER, a multiplayer quiz platform: generating MCQ distractors with sentence transformers, measuring semantic similarity, and examining quiz behaviour for patterns.
Python · Sentence Transformers · AWS · Docker
Earlier work & leadership
Earlier internship
Project Intern
C-DAC Bengaluru
Prototyped a machine-learning framework designed to plug into high-performance computing systems.
Dec 2020 — Jul 2022
Core team → Team Lead
Google Developer Student Clubs · CMRIT
Led the CMRIT chapter after serving on its core team. I was selected as one of 280 GDSC leads in India.
Medium archive
Older notes on data science and machine learning.