"Evidence over opinion. Rubrics before vibes."
Software Engineer with 2+ years shipping production systems on Node.js, Next.js, TypeScript and AWS. For the last year I have worked on LLM evaluation infrastructure: right now the evaluation harness, Docker execution layer, SFT data pipelines and graders behind a Microsoft frontier-model program at Quess Corp (client: Ethara AI). Before that, the internal LLM benchmarking platform at WhatBytes.
How I work: evidence over opinion, boring infrastructure that stays up, and rubrics before vibes. I write the measurement first, then the code.
| 🎯 Pick your lens the site reorders itself for recruiters, engineers or founders |
🤖 Ask Prince an assistant that answers from my real work history |
📄 JD fit report paste a job description, get fit, gaps and a pitch |
✂️ Tailored resume my resume reordered for your role, as a shareable link |
| Docker images shipped for LLM evaluation | open-source repos wired into the eval harness | cut from an AWS bill | concurrent submissions handled by SafeReport | AI response latency after tuning |
Quess Corp · Software Engineer - Gen AI & Full Stack · Feb 2026 – Present · Gurugram, India · On-site · **Client: Ethara AI**
- 🧠 Architected and owned an end-to-end code-migration evaluation harness for a Microsoft frontier-model program, extending the RepoTransBench research benchmark into a production evaluation workflow adopted by the tasking team.
- 🐳 Engineered 1,200+ multi-architecture Docker images (amd64, arm64) across 16 programming languages for LLM evaluation infrastructure, parallelised builds with Docker BuildKit and published to AWS ECR for high-throughput evaluation.
- 🧪 Built an end-to-end Python data pipeline generating a Supervised Fine-Tuning (SFT) dataset for a frontier coding model: mined bounty-linked GitHub issues, filtered issue-resolving pull requests, extracted verified fix and test patches, and produced structured JSONL ground-truth instances across 16 languages.
- 🗂️ Designed a registry-driven evaluation harness with per-repository build, dependency and test configurations for 50+ open-source projects (Redis, curl, systemd, Valkey, DiceDB), enabling deterministic, automated scoring of LLM-generated patches in isolated containers.
- ⚖️ Owned grading and output verification: produced specification-compliant output bundles and agent trajectories, audited every sample end to end, and ran binary-rubric and side-by-side (SxS) evaluations against GPT-5.2 and Gemini-3 Pro, delivering structured JSON reports used in post-training analysis.
WhatBytes · Full Stack Developer · Aug 2025 – Jan 2026 · Bengaluru, India · Remote
- 🤖 Architected an internal LLM benchmarking platform using Next.js, Node.js, and Redis to evaluate and compare outputs across 5+ frontier LLM APIs simultaneously.
- 🔄 Designed multi-turn conversation evaluation pipelines, capturing per-turn scoring and semantic quality metrics across 10,000+ benchmark prompts.
- 🔍 Automated data validation workflows by enforcing structured JSONL output schemas and mandatory rationale checks, cutting manual QC audit overhead by 35%.
- ⚙️ Engineered configurable task-generation modules, allowing engineering teams to dynamically test model outputs against custom rubric criteria and export validated datasets for fine-tuning.
DevTown · SDE I · Jun 2024 – Jul 2025 · Mumbai, India · Remote
- 🔌 Engineered 30+ scalable RESTful APIs using the MERN stack to power core platform features, standardizing frontend-backend data flow and implementing secure JWT-based user authentication.
- 🤖 Architected an AI-powered chat application from the ground up, streamlining component rendering and backend data processing to improve frontend efficiency by 60% and backend performance by 30%.
- ⚡ Optimized AI model response pipelines through strategic API design and advanced caching mechanisms, successfully reducing AI response latency from 1.86s to 1.2s (a 35% improvement).
- 💰 Led a cloud infrastructure cost-optimization initiative, identifying underutilized AWS resources and enforcing S3 lifecycle expiration policies to reduce annual AWS expenditures by over $50,000.
- 🛡️ Executed zero-downtime database migrations (leveraging NeonDB) and implemented robust security protocols, improving overall system scalability and operational efficiency by 40%.
- 🚀 Spearheaded the rollout of new platform features across the MERN stack, establishing reusable UI components that cut down average development time by 40% and accelerated product iteration cycles.
Xipper · Full Stack Developer (Internship) · Nov 2023 – Jan 2024 · Bengaluru, India · Remote
- 🔌 Designed and integrated 25+ RESTful APIs, reducing system overload and enhancing communication between frontend and backend systems by 30%, ensuring robust and scalable interactions.
- 📡 Designed scalable real-time communication features using WebSocket, ensuring smooth and reliable user experiences in live chat applications, reducing latency issues by 35%.
- 🛠️ Participated in regular code reviews and unit testing, identifying architectural bottlenecks and reducing legacy technical debt across shared modules by 30%.
- ⚡ Optimized frontend rendering performance in React.js, reducing unnecessary DOM updates to deliver smoother tab switching and faster client-side navigation.
SafeReport · AI-powered anonymous crime reporting platform · Next.js TypeScript Prisma PostgreSQL Gemini API Docker
- AI-powered incident analysis, automatic report generation and real-time geolocation via the Google Maps API. A complete report in under a minute, with the reporter's identity never stored.
- Prisma/PostgreSQL schema tuned with indexes and query patterns to sustain 10,000+ concurrent submissions, secured with end-to-end encryption, JWT authentication and rate limiting.
TypeScript |
JavaScript |
Python |
Node.js |
Express |
Next.js |
React |
Angular |
Tailwind |
PostgreSQL |
MongoDB |
Redis |
Prisma |
GraphQL |
Supabase |
Firebase |
Redux |
Jest |
AWS |
Docker |
Kubernetes |
GitHub Actions |
Nginx |
Vercel |
Git |
Postman |
Figma |
14 badges · 200-day streak · annual 2024
Fastest reply: email or the contact form on prince404.in. Usually within a day.