Skip to content
View todkarsant's full-sized avatar
🎯
My repositories are no longer centered on models. They're centered on systems.
🎯
My repositories are no longer centered on models. They're centered on systems.

Block or report todkarsant

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
todkarsant/README.md
AI Engineer, multi-agent systems, document intelligence, LLM evaluation, RAG

Dr. Santosh Todkar

I build AI systems that replace manual enterprise workflows — from document intelligence to conversational analytics.

LinkedIn Email Google Scholar Bangalore, India


About

AI Engineer in Bangalore, shipping production agentic AI end to end: multi-agent orchestration, document intelligence pipelines, LLM evaluation platforms, and enterprise RAG — with observability and guardrails built in from day one.

2 enterprise AI systems in production · 7-agent orchestration framework · 6-stage document extraction pipeline · 400+ research citations

Currently researching evidence-to-action reliability in AI agents — measuring whether agents that reason correctly also act correctly.


Open source contribution

Approved

Project What it is Fix
rocm-libraries ★470 AMD ROCm GPU libraries hipDNN Python bindings lacked Graph lifecycle and plan APIs

In review

Project What it is Fix
docling ★70k Document conversion SDK for gen AI Outlook .msg files with non-Unicode bodies crashed the email backend
litellm ★63k AI gateway for 100+ LLM APIs Bedrock Nova Canvas missing-image error surfaced as retryable 500 instead of 400
litellm ★63k AI gateway for 100+ LLM APIs File endpoints raised ValueError (500) for unsupported providers instead of BadRequestError (400)
litellm ★63k AI gateway for 100+ LLM APIs Unanchored regex in duration_parser silently parsed invalid strings as durations
AgentLab ★668 Web agent benchmarking framework No carbon/energy tracking for LLM calls
ou-generativeai-pro ★187 Oracle University GenAI labs Runtime-breaking bugs across agentic AI lab modules
garak ★10k LLM vulnerability scanner Homoglyph detector needed migration to ModernBERTRefusal
openlit ★3.1k AI observability and cost optimization Qdrant instrumentation dropped span attributes for positional args
openlit ★3.1k AI observability and cost optimization CLI --no_capture_message_content option was declared but never registered in argparse

Projects

Trade Finance Document Intelligence

Designed a multi-stage document extraction workflow for Letters of Credit, combining PDF processing, section-aware chunking, retrieval and structured field extraction with validation. Human reviewers now verify AI output instead of doing the work themselves.

View project

SWARM — Smart Workflow Agents for Reporting & Metrics

A multi-agent conversational system — 7 specialist agents orchestrated by LangGraph — that translates plain English questions into live analytics with charts and tables. Users ask "Show me EBITDA trend for Q3" and get answers in seconds, not days.

View project

AI Agent Evidence-to-Action Reliability

A research framework investigating whether agents that interpret evidence correctly also select and execute appropriate actions. Quantifies the gap between "understanding" and "doing" that most evaluations miss.

View project

LLM Evaluation and Observability Platform

A reproducible evaluation platform that detects regressions across model versions, prompt variants, and retrieval strategies — catching quality drift before it reaches users.

View project


Toolkit

LangGraph Azure OpenAI Model Context Protocol RAG and Vector Databases Python, FastAPI, Docker, PostgreSQL, Redis, Azure, Git

Pinned Loading

  1. Enterprise-Analytics-Copilot Enterprise-Analytics-Copilot Public

    Production-oriented local reference implementation for NL2SQL + analytics reasoning. This project is designed as a portfolio-grade Applied AI system rather than a simple `question -> LLM -> SQL` demo.

    Python 1

  2. LLM-Evaluation-Observability-Platform LLM-Evaluation-Observability-Platform Public

    A reproducible evaluation platform that detects quality regressions across model, prompt, retrieval and application versions while correlating quality with latency, token usage and cost.

    Python