Agentic AI for finance.
I build AI agents and ML systems that forecast cash, price risk and automate finance work, backed by a Ph.D. in Statistics and 14 years in lending and capital markets.
$ agent.run("introduce Avinash Barnwal") plan retrieve profile → call tools → summarize tool read_cv() ✓ Ph.D. Statistics, Stony Brook · IIT Kharagpur tool query_experience() ✓ 14 yrs across lending, credit & capital markets tool current_role() ✓ Head of Analytics & Data Science @ Autocash.ai reflect focus = agentic AI for finance done "Builds AI agents and ML systems that help businesses make better money decisions."
Experience across
- Autocash.ai
- Kapitus
- Deutsche Bank
- Indiabulls
- Fullerton India
- Stony Brook
- IIT Kharagpur
Agent at work
An agent running a 13-week cash forecast.
Cash dips to $232k, $18k under the $250k buffer. Recommend moving two vendor payments ($40k) from week 7 to week 9.
About
A statistician building agents.
Trained as a statistician, I found my way through machine learning to build models that change real decisions. Today I work on AI agents that plan, use tools and reason over financial data, so businesses, especially small ones, can forecast cash, manage risk and get better financing.
- 14+
- years in finance & data
- Ph.D.
- Statistics, Stony Brook
- 4
- research publications
- 6
- companies, banks & fintechs
What I build
From models to autonomous workflows.
Agentic finance workflows
LLM agents that plan, call tools and check their own work across cash forecasting, treasury and AR/AP.
Credit & risk ML
Predictive models for small-business lending: underwriting, risk and portfolio decisions in production.
ML that ships
Evaluation, monitoring and quality systems that get models out of notebooks and into the business.
How my agents work
Plan. Act. Check. Remember.
- 01
Plan
Break a finance goal, like a 13-week cash forecast, into concrete steps.
- 02
Act
Call tools: ledgers, bank feeds, models and SQL, instead of guessing.
- 03
Check
Reconcile results against the books and flag exceptions for a human.
- 04
Remember
Keep context across runs so every cycle is sharper than the last.
Try it
Ask my agent.
A small demo of the pattern: each question is routed to a tool, and the answer comes back with links. It's scripted and runs entirely in your browser, with no model calls.
Writing
Notes on AI & ML.
Building a Foundation Model for Personalized Recommendation
One model. Every member’s full history. Many recommenders.
Causal Reasoning for Agentic AI
An agent I watched last quarter did everything right and still got the answer wrong. It read the request, produced a ...
When We Can Finally See Inside an AI’s Mind
Why understanding how large language models think may become one of the most important breakthroughs in artificial in...