Ameya Raj on Tal
Headline
Software Engineer @ One (former Pica) | Systems, Algorithms, Automations
About
Software Engineer who thinks in systems, not features. I build platforms that coordinate complex workflows — dependency graphs, distributed state machines, self-improving execution pipelines. The problems I enjoy most are the ones where being almost right is worse than being obviously wrong: race conditions across distributed pods, database scoping bugs that silently corrupt historical data, or getting 12 out of 12 regime clusters to actually activate in a self-supervised model. What I'm working on: At Pica, I built an autonomous API testing platform that validates 200+ platform integrations. The interesting parts: topological sorting for action ordering, interrupt-resume logic that persists full execution state across pod restarts, and a two-pass system where the AI agent learns from its own failures to retry intelligently — 70% cost reduction, 80% higher success rate. On the side, I'm building Quantara — a self-supervised financial foundation model (84M params, scaling to 350M) that learns representations across 5 markets (US equities, Indian equities, crypto, commodities, forex). Factorized attention architecture, 6 training objectives, multimodal news fusion with 29M+ GDELT events, all engineered to train on free-tier T4/P100 GPUs with fp16 and gradient checkpointing. Previously: data processing pipelines at Dom Ventas (90% reduction in report generation time, 500+ users), survey platform at Tongmyong University (international collaboration, 1000+ concurrent users). Education: BS in Data Science from IIT Madras, BSc in Computer Science from University of Delhi. Tech I reach for: Python, TypeScript, C++, PyTorch, NestJS, PostgreSQL, Docker, Polars. But I care more about the problem than the tool. Always interested in connecting with people who build systems where correctness matters.
Work experience
Software Engineer at One (formerly Pica), Apr 2025 - Present · 1 yr 6 mos, San Francisco Bay Area
Full Stack SDE Intern at Dom Ventas, Jun 2024 - Nov 2024 · 6 mos, Bengaluru, Karnataka, India, ❖Platform Development & Security: Developed an Amazon Ads optimization web app using Python Flask, Vue.js, Vuetify, and MySQL. Implemented secure user authentication (username/password, Google, Amazon) and a tiered subscription model (Basic, Premium, Elite). ❖Data Processing & Reporting: Built tools for Excel-based campaign data processing and automated report generation (tracking impressions, clicks, conversion rates) with asynchronous API calls that reduced report time by over 90%. ❖Integrations & Deployment: Integrated Razorpay for payments, enforced RBAC for secure access, set up email automation with Celery, and deployed the platform on a scalable Linux VPS.
Web Developer at Tongmyong University, Jun 2022 - Oct 2022 · 5 mos, South Korea, Connected with people from South Korea and worked in a team to Develop a Survey website using PHP, html and MySQL. Made Signup/Login, Home, Questions and End Page. Used session to store/retrieve portions of already filled form. Created Database schema.
Education
Bachelor of Science - BS, Data Science and Programming, Indian Institute of Technology, Madras
B.Sc, Computer Science, PGDAV College
Public links
github: https://github.com/ameyarj
