Software Engineer - AI in Mesa, Arizona
I build AI products and backend systems that solve real problems for real people. I start with the person who has the problem, then work backward to the smallest reliable system that fixes it.
You can run my work, not just read about it: akashagrawal.dev
Regalia Pass On
Graduation regalia costs a lot, gets worn once, then sits in a closet while the next class buys new. I am building a marketplace so students can pass theirs on. Launching soon.
ReFocus.AI
Most focus apps just block distractions. They never help you notice you drifted or get back on track. I built the prototype in a 24-hour hackathon, believed in the idea, and I am rebuilding it for a public launch. See the hackathon project.
NewsGenie · Software Engineer - AI · Feb 2026 to now
Publishers were losing articles to scrapers faster than they could respond. I built a service that encrypts content per reader and per article, and cuts off access in under a second across separate services. It blocks 7 categories of bots. I also built an agentic LangChain pipeline that removed over 1,000 hours of manual editorial work a year.
Arizona State University · AI Engineer · Aug 2024 to Jul 2025
Researchers were tracking soil particles by hand, about 40 hours per experiment. I replaced that with a serverless pipeline on AWS Lambda and a tracker built on YOLOv8 and FairMOT. A cycle now takes 2 hours, holds identity on 3,500+ particles at 0.82 mAP, and stayed up 99.9% of the time.
LTIMindtree, for CITI Bank · Software Engineer · Jan 2023 to Nov 2023
Engineers changed financial system configuration by hand, which is slow and leaves no audit trail. I built Node.js and Angular middleware with single sign-on, role-based access, and full audit logging, and made data processing 35% faster.
LTIMindtree, for Avery Dennison · Software Engineer - Data · Aug 2021 to Dec 2022
The same customer existed many times over across regional systems, so no report could be trusted. I built an Informatica pipeline that consolidated 10 million records into 2.5 million golden records, improved data governance 4x, and made the legacy pipeline 30% more efficient.
| Area | Tools |
|---|---|
| Languages | Python, TypeScript, JavaScript, Java, SQL, C++, Bash |
| AI and ML | LangChain, LangGraph, RAG, multi-LLM orchestration, PyTorch, HuggingFace, OpenCV, RAGAS |
| Backend | FastAPI, Node.js, Next.js, React, REST, GraphQL, Spring Boot |
| Cloud and infra | AWS Lambda, S3, KMS, CloudWatch, Docker, Kubernetes, Terraform, Jenkins |
| Data | PostgreSQL, Oracle, MongoDB, Redis, ChromaDB, Informatica ETL |
- LambdaLens analyzes video on AWS Lambda and cut its own cloud bill in half.
- Portfolio runs a SQL engine I wrote from scratch, plus real encryption, in your browser.
- TrendScout AI and AI Agent Chatbot are where I test agent ideas.
- NeetCode submissions keeps my problem solving sharp. My ICPC team placed 239th of 10,000 in 2019.
I am looking for teams building AI products that people depend on daily. I am most useful when a product needs someone who can own it end to end: talk to users, design the system, ship it, then measure whether it actually helped.
I care about the unglamorous parts. Retries that recover, keys that stay protected, pipelines you can re-run, and numbers that hold up when someone checks them.
Happy to talk about AI systems, backend architecture, or what to hike in Arizona.
M.S. Computer Science, Arizona State University, December 2025