profile-pic

Madhav Mittal

I’m a Machine Learning Engineer focused on building applied LLM systems that are useful in real workflows, not just demos.

My work sits at the intersection of ML engineering, product thinking, and systems design: retrieval pipelines, agent orchestration, reasoning loops, evaluation, and developer tooling. I’m especially interested in AI systems that help engineers and operators work better by automating parts of complex workflows while keeping humans in the loop.

I currently work at Visa, and previously worked across quantitative research and engineering roles at WorldQuant and IIT (BHU) Varanasi. Those experiences pushed me toward solving messy, real-world problems with a mix of modeling, software, and practical iteration.

I studied Mathematics and Computing at IIT (BHU), where I developed a strong bias toward difficult problems, first-principles thinking, and building systems that hold up outside toy settings.

I’m particularly excited by teams working on applied AI, LLM infrastructure, agentic systems, developer tools, and vertical workflow automation. Always happy to connect with people building ambitious AI products.

  • Role

    Forward Deploy AI Engineer

  • Years of Experience

    5.1 years

Skillsets

  • LLM
  • Vector databases
  • TypeScript
  • Statistical Modeling
  • SQL
  • Retrieval
  • Rest APIs
  • rag
  • Python
  • Predictive Modeling
  • PostgreSQL
  • Optimization
  • Openclaw
  • Neo4j
  • MCP
  • Agents
  • LangSmith
  • LangGraph
  • Java
  • Git
  • Feature Engineering
  • FastAPI
  • Docker
  • CI/CD
  • C++
  • BigQuery
  • Bash
  • backtesting
  • AWS

Professional Summary

5.1Years
  • Jun, 2024 - Present2 yr

    AI Engineer

    Visa
  • Aug, 2023 - Present2 yr 10 months

    Research Consultant

    WorldQuant
  • May, 2022 - Oct, 2022 5 months

    Compiler Developer

    Python Software Foundation

Applications & Tools Known

  • icon-tool

    TypeScript

  • icon-tool

    SQL Server Reporting Services

  • icon-tool

    REST API

  • icon-tool

    Python

  • icon-tool

    PostgreSQL

  • icon-tool

    Neo4j

  • icon-tool

    Java

  • icon-tool

    Git

  • icon-tool

    FastAPI

  • icon-tool

    Docker

  • icon-tool

    C++

  • icon-tool

    BigQuery

  • icon-tool

    Bash

  • icon-tool

    AWS (Amazon Web Services)

Work History

5.1Years

AI Engineer

Visa
Jun, 2024 - Present2 yr
    Built and deployed R2D2, an internal AI system that converts Jira requirements into code drafts, test scaffolds, and CI-ready pull requests, automating large parts of the development workflow. Designed a repository-aware retrieval pipeline to fetch relevant code, reducing invalid and out-of-context suggestions. Developed evaluation and CI validation pipelines with clear quality metrics for AI-generated code; drove adoption across 800+ engineers and received the Visa Innovator Award. Designed a unified data model integrating raw campaign data into a centralized analytics layer. Built agent-driven workflows that converted fragmented analytics data into structured diagnostics and campaign performance projections. Led AI-driven automation of NexusIQ vulnerability triage and dependency remediation workflows across 20+ engineering teams, coordinating safe package version upgrades. Contributed to backend services using Java and REST APIs; built geo-targeted segmentation proof-of-concepts used in experimentation.

Research Consultant

WorldQuant
Aug, 2023 - Present2 yr 10 months
    Developed predictive alphas using statistical feature engineering, systematic backtesting, and out-of-sample validation across diverse market regimes. Built evaluation pipelines for signal stability, drawdown analysis, and risk-adjusted performance, reducing overfitting and improving robustness.

Compiler Developer

Python Software Foundation
May, 2022 - Oct, 2022 5 months
    Extended the LPython compiler by implementing intrinsic math and statistical modules, modifying AST and backend components in the core compiler pipeline. Built compile-time and runtime test harnesses that improved correctness and system stability across language features. Contributed intrinsic integrations including loop and array constructs, expanding LPython's numerical computing capabilities.

Major Projects

2Projects

Coviction AI Agentic VC Memory & Conviction Intelligence Platform

Jan, 2026 - Dec, 2026 11 months
    Built Coviction AI, an AI-native venture capital memory platform that turns fragmented VC notes, calls, founder conversations, and deal signals into a persistent knowledge layer. Architected the intelligence pipeline across structured LLM extraction, retrieval-augmented reasoning, graph-based entity linking, Bayesian-style belief updates, temporal signal decay, sentiment scoring, semantic search, and thesis synthesis. Developed and deployed the full-stack MVP using FastAPI, PostgreSQL, async AI workflows, multimodal ingestion, knowledge graph visualization, automated briefs, conviction scoring, voice/image capture, and cloud deployment.

StrataAlpha Multi-Agent Thesis-to-Trade Engine

Jan, 2026 - Dec, 2026 11 months
    Built StrataAlpha, a multi-agent equity research engine that transforms natural-language investment theses and live market data into structured, confidence-weighted Buy/Watch/Defer decisions using retrieval-augmented reasoning, factor modeling, and thesis-aware analysis workflows. Implemented an architecture spanning thesis decomposition, evidence gathering, factor/risk modeling, thesis synthesis, and action routing with explicit tool contracts, guardrails, and per-agent evaluation metrics. Self-hosted the core LLM stack and deployed StrataAlpha with authentication, sessioned workflows, and shareable research memos.

Education

  • Dual Degree (B.Tech + M.Tech)(Hons) in Mathematics and Computing

    Indian Institute of Technology (BHU) (2024)