Skip to content

Dhairya Mishra · Software + ML Engineer · New York

I turn ideas into reliable software products.

Dhairya Mishra

Former Senior Software Development Engineer at CVS Health. I build the systems around production AI: the product experience, APIs, evaluation, infrastructure, and operational safeguards that make the work useful.

4+ years shipping software and ML systems Enterprise + startup operating range End-to-end technical ownership

How I build

The work lives at the intersection of product, systems, and applied ML.

I am most useful when a team needs more than a model or a feature: a dependable system that fits real workflows, makes deliberate tradeoffs, and has a clear outcome.

Product delivery

React, TypeScript, FastAPI, Node.js, and typed API contracts for systems people can actually use.

Applied ML

RAG, computer vision, LLM-as-judge evaluation, model serving, and reproducible inference workflows.

Reliable systems

Cloud infrastructure, observability, testing, deployment, and the operational boundaries around production models.

Enterprise impact

Evidence, not adjectives.

The visual work shows technical range. These production examples show what changed for real teams and workflows.

01

Enterprise AI system

Made recurring engineering support questions self-service.

Problem

Engineering teams were repeatedly routing common troubleshooting questions through manual support instead of drawing on existing ticket history and documentation.

What I owned

Built the self-service retrieval and answer workflow over ticket history and internal documentation.

Result

16% ticket deflection · 20% faster resolution

RAGTicket historyInternal docsEnterprise support
Read the RAG assistant case study

02

Applied ML workflow

Put a quality gate between model output and production content.

Problem

Catalog imagery needed better captions, but automatically replacing existing copy without a quality boundary would create brand and reliability risk.

What I owned

Built a vision-language generation pipeline and an LLM evaluation loop that only wrote back captions when they beat the existing text.

Result

18% catalog coverage lift

VLMsLLM evaluationPythonCI/CD
Read the VLM workflow case study

Career signal

Built in environments where reliability, users, and business outcomes all matter.

My experience includes customer-facing platform reliability, internal developer tooling, enterprise applied AI, and startup experimentation. The common thread is taking ambiguous work through to a system people can rely on.

See the full experience timeline
CVS Health

Sr. Software Development Engineer

CVS Health · 2024–2025

Evidenza

AI Engineer Intern

Evidenza · 2026

Aetna

Advanced Software Developer

Aetna Health · 2022–2023

Start here

Looking for someone who can own both the model and the system around it?

I am open to conversations about software engineering, applied ML, platform work, and product teams shipping AI responsibly.