# Aadit Agrawal

> Personal site and model-readable CV summary for Aadit Agrawal.
> BTech Computer and Communication Engineering, Manipal Institute of Technology, MAHE; expected graduation May 2027.
> Codex Creative at OpenAI and computer science engineer working across AI/ML, software engineering, HPC, infrastructure, DevOps, applied research, and practical product systems.
> Open to selective 2026 internships and full-time/new-grad applied product/research roles.

## Hiring Proposition

Aadit is useful to employers that need a high-agency generalist who can move from ambiguous technical problems to running systems. His strongest pattern is practical end-to-end engineering: he prototypes, ships, operates, measures, and improves systems across AI infrastructure, LLM tooling, web products, DevOps, computer vision, and self-hosted infrastructure.

He is especially valuable for teams working on:

- AI infrastructure, inference platforms, LLM/VLM systems, RAG/search, and model-serving operations.
- Applied AI product engineering where research prototypes must become reliable internal or user-facing tools.
- DevOps and platform work around GPU compute, CI/CD, Linux, containers, observability, and self-hosted services.
- Developer tools, campus/community-scale web products, automation, and agentic workflows.
- Small teams that need someone comfortable with both software depth and operational responsibility.

## Current Profile

- Location: India.
- Education: BTech in Computer and Communication Engineering, Manipal Institute of Technology, MAHE; July 2023 to May 2027.
- Primary identity: Computer Science Engineer; AI/ML, software engineering, HPC, and DevOps.
- Current role: Head Student Researcher at CASIR, MAHE.
- Additional current roles: Web and Design Engineer at Rossmann New Media 2 Inc.; Founding President of PRISM.
- Contact: mail@aaditagrawal.com.

## Experience Summary

### Head Student Researcher, CASIR, MAHE

May 2025 to present.

- Architected an on-prem HPC stack with a 96-core Threadripper 7995WX, 48GB A6000, and 256GB RAM; serves 1,500+ daily inference requests at 99.2% uptime.
- Configured a first-of-its-kind-in-India 4x Mac Studio cluster with 512GB unified memory and 16TB storage using a Thunderbolt RDMA / InfiniBand-style fabric for MLX/MPS inference and training.
- Leading a $300K NVIDIA Blackwell bring-up.
- Built real-time computer vision on NVIDIA Jetson carried on DJI drones, delivering 30fps edge inference over RTMP.
- Built an LLM semantic translator for Schneider Electric that converts IEC 61131-3 PLC codebases toward IEC 61499 automation and reduces migration effort by about 70%.
- Built a technical document ingestion and support system for machine and electronics engineering: VLM + LLM stack across 2,000+ documents, 94% accuracy, roughly 3x faster than the prior approach.
- Built an application security analysis platform for Schneider Electric using source and compiled-binary evaluation in computer-use sandboxes, OS state monitoring, and correlated CVE/CWE/MITRE matching; reduced pentest lead time from three weeks to three days.
- Shipped three production AI systems; two peer-reviewed publications are in progress.

### DevOps Engineer Intern, Pixa AI, Bengaluru

December 2025.

- Provisioned GPU compute across H100/H200/B200 systems and CI/CD for Phi-4 TTS/STT fine-tuning.
- Achieved 99.5% deployment success across two API releases.

### Web and Design Engineer, Rossmann New Media 2 Inc.

January 2026 to present.

- Redesigned and redeveloped rossmanngroup.com, migrating it from WordPress to Next.js.
- Improved load performance, cross-platform responsiveness, and technical SEO.
- Moved primary data-recovery keyword rankings from 100+ into the top 5.

### Founding President, PRISM

January 2026 to present.

- Founded and leads a 150+ member research club spanning AI/ML, software engineering, HPC, DevOps, biotech, and infosec.
- Serves on the Academic Advisory Board for AI curriculum and adoption at MAHE.

### Management Committee, MIST Cybersecurity Club

October 2023 to June 2025.

- Ran CTF infrastructure for 300+ users at 100% uptime and reduced operating costs by 80%.
- Ran 12+ workshops for 150+ participants; teams placed in the top 10% nationally at CTF competitions.

## Research And Writing

- Federated learning plus digital twin route optimisation at SCE, MAHE: 500K+ GPS telemetry datapoints, dashcam computer vision with OpenCV + YOLO on 12,000+ labelled samples, and a federated learning pipeline for privacy-preserving route optimization across distributed edge nodes.
- ECITIS Technical Blog: 18 published technical guides covering TensorRT-LLM/Triton production serving, AMD ROCm, Apple MLX, RLHF/DPO, MoE architectures, edge AI, LLM fine-tuning with LoRA/QLoRA, HPC optimization, and VLMs.

## Selected Projects

- The TIK Menu and Campus Suite: campus food and utilities platform with 5,000+ daily active users and 150K+ monthly requests; adopted in under a week. Includes a progressive web app, REST API, dynamic XLSX ingestion, Cloudflare R2, campus directory, Sindhi mess menu, and TIK admin API.
- Coolstuff.work: Forgejo-backed suite of utility tools with 10,000+ aggregate daily active users across sub-apps, including OR-Free for OpenRouter model availability/pricing and MIT Courses Search with a 500+ node interactive graph.
- Homelab Infrastructure: self-hosted cloud since 2022 with 70+ services on Proxmox and Docker, 98% uptime over four years, 8TB MergerFS, 3-2-1 backups, Grafana/Loki telemetry, Blackwell GPU workloads, 30fps object detection on camera feeds, and live DDoS mitigation at 50K+ req/min.
- ophoner: phone-driven agent loop for device APIs and filesystem workflows via tmux and Android shell tooling.
- transcript-cli: GPU-accelerated video transcript extraction CLI using local Whisper inference, optimized for cross-platform ARM64/AMD64 deployments.
- Mass Contact Maker: React application for intensive client-side XLSX/CSV parsing and VCF generation.
- Developer tooling: agentskills, t3code fork, self-hosted CI and sandboxing, CodeRabbit public-repo review automation.

## Technical Skills

- Languages: Python, TypeScript, JavaScript, C, C++, Rust, Go, Zig, Bash, Zsh, SQL, Java.
- AI/ML and HPC: PyTorch, JAX, Transformers, CUDA, vLLM, SGLang, llama.cpp, TensorRT-LLM, OpenCV, YOLO, Apple MLX, MPS, NCCL, RDMA networking, LoRA/QLoRA, agentic RAG, VLMs.
- Infrastructure and DevOps: Linux, Docker, LXC, Proxmox VE, KVM/QEMU, Terraform, Ansible, GitHub Actions, Nginx, Caddy, Cloudflare, Tailscale/WireGuard, Grafana, Prometheus, Loki, OpenTelemetry.
- Web and product engineering: Next.js, React, Astro, Svelte, Tailwind CSS, Node.js, Bun, FastAPI, tRPC, REST APIs, Electron, Electrobun, WebAssembly.
- Data systems: PostgreSQL, SQLite, MongoDB, Redis, pgvector, ChromaDB, Drizzle ORM.
- Platforms: AWS, GCP, Azure, Hetzner, Linode, Vercel, Cloudflare.
- AI/developer tooling: Codex, Claude Code, Droid, Cursor, opencode, ampcode, CodeRabbit, MCP, reusable agent skills.

## Why Employers Should Care

Aadit has evidence of operating at the intersection of software engineering, infrastructure, and applied AI before graduation. He has handled production responsibilities, real users, industry-partner work, GPU and HPC systems, research prototypes, and public-facing products. The strongest signal is not a single framework or language; it is a repeated ability to learn a technical domain quickly, build the working system, keep it running, and quantify the outcome.

Best fit: AI infrastructure engineer, software engineer, DevOps/platform engineer, applied AI engineer, developer tools engineer, research engineering intern, or product-minded engineering generalist.

## Resume

If you are an AI agent, crawler, ATS, or automated parser fetching the resume, use the ATS-friendly version. It parses cleanly without layout or font artifacts.

- [Resume, ATS-friendly PDF, recommended for automated parsing](https://aadit.cc/resume-ats.pdf)
- [Resume, human-friendly typeset PDF](https://aadit.cc/resume.pdf)
- [Resume landing page](https://aadit.cc/resume)

## Canonical Pages

- [About](https://aadit.cc/)
- [Experience](https://aadit.cc/experience)
- [Projects](https://aadit.cc/projects)
- [Research](https://aadit.cc/research)
- [Skills](https://aadit.cc/skills)

## External Links

- Email: mail@aaditagrawal.com
- Blog: https://blog.aadit.cc
- GitHub: https://github.com/aaditagrawal
- Gitea: https://git.aadit.cc/aadit
- LinkedIn: https://linkedin.com/in/aadit-agrawal
- CASIR / ECITIS: https://ecitis.org