SDE I @ Amazon · New York
I'm Tejas—an SDE at Amazon and Georgia Tech alumnus working across production software, machine learning, and quantitative finance.
01 / Selected work
Selected products and experiments across software, data, and finance.
Illustrative comparison
No scenario values shown · try the synthetic demo
Sell vs SBLOC decision model
A full-stack simulator that compares selling equities with borrowing against them for a real-estate down payment, modeling taxes, opportunity cost, cash flow, and margin-call risk.
Three-pass aggregate · 19 questions
Baseline → planner · measured aggregate, limited scope
Planner and evaluation harness
Across three passes on 19 questions, the planner reported a 4.04 vs 3.93 mean judge score, 0.580 vs 0.579 keyword recall, and 23.3s vs 14.7s mean request latency.
What the projects address, my contribution, and their potential value. Request a walkthrough to continue the conversation.
Personal finance workspace
Source code is private.
A personal finance application that brings accounts, transactions, budgets, and liquidity planning into one workspace.
Personal AI coordination
Source code is private.
A personal AI coordination system that manages assistant requests, permissions, and human approvals.
Market research and personal watchlists
Source code is private.
A market-research workspace with quotes, news, and personal watchlists for signed-in users.
Personal AI chat interface
Source code is private.
A private chat interface for a personal AI system, with conversation history, model choices, and approval prompts.
Inactive research project
Source code is private · Inactive project.
An inactive research bot for comparing related prediction markets across exchanges.
02 / Experience
Ambitious systems, clear ownership, work that holds up in production.
Amazon · New York, NY
SDE I in New York. Converted from the internship that shipped a Bedrock data-retrieval agent for Creator Assistant.
VIVA Finance · Atlanta, GA
Modernized loan-servicing workflows spanning forbearances, settlements, principal changes, and reimbursements.
Amazon · New York, NY
Built a data-retrieval agent for Creator Assistant with Amazon Bedrock, Java, Spring Boot, and internal reporting APIs. Used Return Control so tool calls execute in the application rather than the model, and automated evals for accuracy, relevance, latency, and tool selection.
GT Investments Management Club · Atlanta, GA
Founded and led a student-managed fund with $100k+ in privately raised capital. Authored the investment strategy, managed sector leads, and built quantitative tools for the team.
CVS Health · New York, NY
Implemented automatic client-side field-level encryption for sensitive patient data in a Spring Boot and MongoDB platform.
03 / About
I like turning messy questions into things people can actually use.
I earned both my Master's in Machine Learning and B.S. in Computer Science from Georgia Tech. My graduate research explored Proofs of Useful Work through arbitrary matrix multiplication; my product work spans agentic AI, secure data systems, and quantitative tools.
Away from the terminal, I follow markets, tinker with local AI hardware, and get outside when I can.
What I build with
Languages, ML, and systems tooling I use regularly.How I use AI
Local models, real hardware, agentic tooling.04 / The personal feed
A live feed of activity.
Run · Oct 3
Sport · Oct 3
Current direction
05 / Say hello
For a technical walkthrough, include the project name in your note.
Request a technical walkthrough