What does Bhavya Dhoot actually do?

He works as a forward deployed engineer: he embeds with a business, finds where it is losing time or money, and ships the system that closes it. That has spanned finance reconciliation, sales reporting, supply-chain document extraction, production planning and after-hours support, alongside pricing and risk tooling on the quantitative side.

Is Bhavya Dhoot available for hire or consulting?

Yes. He is open to quantitative developer and AI systems engineering roles, and to AI automation consulting engagements. He is a computer science undergraduate at VIT graduating in 2027 and currently does quant research at JM Financial.

What is Bhavya Dhoot's strongest quantitative result?

A dual-engine NSE momentum system with a 1.44 out-of-sample Sharpe ratio and a 6.6 percent maximum drawdown.

Can I try his work rather than read about it?

Yes. The home page runs four working tools in the browser: a two-pass reconciliation matcher, a five-input market signal engine, a Black-Scholes payoff and Greeks explorer, and an implied-volatility solver benchmark. Every figure is computed in the visitor's browser.

Why is the Health Factory work described vaguely?

That engagement is under NDA. Bhavya describes the class of problem and the engineering approach, never the client's architecture or data.

Deployed systems,
not
prototypes.

I'm a CS undergraduate at VIT (class of 2027) working as a hands-on AI consultant and quantitative developer. From a founder's office, I build and operate 7+ production AI and automation systems across two FMCG companies, covering sales intelligence, production planning, and supply-chain reconciliation.

Behind that sits quantitative research at JM Financial: fine-tuning a foundation model on NSE equity data and building an options pricing and paper-trading terminal. Every system I ship is judged on one question: does it keep running, correctly, without me in the loop?

2027 B.Tech CS, VIT · Expected

6 Institutions Shipped For

139 Tests on Flagship Engine

Operating History

Systems shipped
across institutions.

Equity Research Intern

JM Financial

2026 Derivatives · Market Microstructure

Built quantitative research and trading infrastructure for Indian equity derivatives, from a fine-tuned foundation model on NSE data to a production-grade options pricing terminal.

→ Fine-tuned a 24.7M-param foundation model feeding a 5-layer signal pipeline, <340ms latency
→ BSM options pricer: warm-start IV solver, 80-90% compute cut in production
→ Broker API quota down 66%; bandwidth down 90%+ with gzip

AI Consultant · Founder's Office

The Health Factory

2026 · Present Production AI · Automation

Own AI and automation infrastructure end-to-end across two sister FMCG companies: scope use cases with leadership, select the stack, then build, deploy, and operate 7+ production systems.

→ Citation-backed sales-intelligence platform: ERP + quick-commerce data in one live dashboard
→ LLM document extraction turning supply-chain paperwork into a same-day ledger
→ Zero-touch daily reporting with delivery confirmation and retries

AI Automation Intern

GoHappyClub

2026 · Present Multi-Agent Systems · RAG

Built and deployed ARAS, an autonomous research pipeline orchestrating 6 specialized AI agents from idea through novelty gate, experiments, and figures to a finished paper.

→ 6-agent autonomous research pipeline, idea to finished paper
→ 88% retrieval accuracy over a 50K+ document corpus (FAISS)
→ 70% cut in manual research cycle time

AI Automation Consultant

Independent

2025 · Present Voice Agents · Workflow Automation

Scope and deliver automation systems for overseas clients, from outbound voice-agent workflows to conversion audits.

→ Voice-agent workflow (n8n + RetellAI) automating order follow-up calls for a US distributor
→ Full CRO audit and 4-vertical outreach library for an AI agency

Research Analyst Intern

Moon Finance

2025 Equity Research · Valuation

Equity research across 50+ securities and 8 sectors, with Python-based portfolio optimization behind real allocation decisions.

→ DCF and comps valuation models, 75% recommendation accuracy
→ Portfolio optimization over 500K+ data points, informing $50K+ allocations
→ 15+ investment reports across 8 sectors

Chairperson

Entrepreneurship Cell · VIT

2026 · Present Leadership · Capital Allocation

Re-elected to lead a 200+ member organization; direct strategic priorities, capital deployment, and cross-functional operations for the 2026 cycle.

→ Rs. 25L+ (~$30K) annual budget across 5 verticals, zero cost overrun
→ Built recruitment and performance frameworks scaling the analyst pipeline

How I think
about systems.

Deterministic core, LLM at the edges

Numbers a founder acts on cannot hallucinate. Insight engines should be deterministic and citation-backed, with LLMs layered on top for narration, never for arithmetic. That is how the sales-intelligence platform is built, and why its answers are auditable.

Homogeneity is a systemic risk

When every trader runs the same model, crashes amplify. My agent-based simulations measured +30.3% drawdown amplification as LLM adoption scaled to full homogeneity. Computational Asymmetry Theory is my framework for pricing that risk.

III

Verification is the edge

A backtest is a claim, not a result. I rebuilt my momentum engine in Rust to cross-validate the Python research code to the decimal, and rejected roughly 20 strategy families before trusting one. Alpha either survives scrutiny or it was never alpha.

Straight answers

The questions
people ask.

What do you actually do?: I work forward deployed: I sit with the people doing the job, find where the business is losing time or money, and ship the system that closes it. That has meant reconciliation in finance, reporting in sales, document extraction in supply chain, demand planning in production and an after-hours voice agent in support — plus pricing and risk tooling on the quant side. The common thread is that it has to keep working correctly after I stop watching it.
Are you available, and for what?: Yes — for quantitative developer and AI systems engineering roles, and for AI automation consulting engagements. I am a computer science undergraduate at VIT graduating in 2027, currently doing quant research at JM Financial, and I have shipped production systems alongside study since 2024. Email is the fastest route.
What is the strongest quantitative result you can show?: A dual-engine NSE momentum system with a 1.44 out-of-sample Sharpe and a 6.6% maximum drawdown. Methodology is available on request; the code and a fuller write-up are linked from the work page.
Can I see something running rather than described?: The home page runs four working tools in your browser: a two-pass reconciliation matcher, a five-input market signal engine, a Black-Scholes payoff and Greeks explorer, and an implied-volatility solver benchmark. Every figure they show is computed on your machine as you scroll, not written into the page.
Why is the Health Factory work described so vaguely?: That engagement is under NDA. I describe the class of problem and the engineering approach, never the client's architecture or data. The reconciliation demo on the home page runs on a fixture invented for this site for exactly that reason.

Systems Architecture

Technical
infrastructure.

Languages & Core

Python · Pandas · NumPy · Polars
Rust · TypeScript / JavaScript
SQL · PostgreSQL · TimescaleDB
MongoDB · Redis · SQLite

Quant & Research

Options Pricing (BSM) · Greeks · SPAN Margin
Cross-Sectional Momentum · Factor Research
Backtesting · Monte Carlo Simulation
GARCH · HMM Regimes · Time Series

AI & ML

LLM Integration (OpenAI · Claude · Gemini)
RAG Pipelines · FAISS · ChromaDB
Multi-Agent Systems · n8n
PyTorch · LightGBM · RL (PPO) · Qiskit

Infrastructure & Ops

FastAPI · Node.js · Next.js · React
Docker · GitHub Actions · systemd
GCP Cloud Run · Oracle Cloud · Render
Playwright · Airflow · MLflow · Odoo ERP

Deployed systems, not prototypes.

Systems shippedacross institutions.