ISHAN BHAT DEVELOPER · SEATTLE, WA

Ishan
Bhat.

Developer building software where backend systems, data, and real-world operations meet.

I enjoy turning complex technical systems into simple, useful tools that people can act on. I work across backend systems, data platforms, and distributed systems, with applied AI as another tool in the toolkit.

GitHub LinkedIn Résumé

Backend · Data · Cloud

DEVELOPMENT / INFRASTRUCTUREBACKEND ↔ APPLICATIONSBACKEND · DATA · CLOUD

Get to know me

About

Developer with experience building backend services, operational data tools, and infrastructure software for engineers.

I'm pursuing an M.S. in Computer Science at Purdue, where I research scheduling and routing for LLM inference. Before that, I completed a Systems Engineer internship at Amazon Prime Air, building signal-analysis algorithms and telemetry tools for drone engineers. At Capital One, I worked on cloud infrastructure and monitoring; earlier, at Cylus, I built real-time software for railway cybersecurity.

U.S. Permanent Resident (Green Card) — authorized to work without sponsorship.

Where I've worked

Experience

Graduate Researcher, Agentic Systems
Jan 2026 – Present
Purdue University · Prof. Vamsi Addanki's Lab— West Lafayette, IN
- Designed an Earliest-Deadline-Late (EDL) scheduling policy for LLM inference serving, improving goodput and deadline adherence for multi-step, latency-sensitive agent workflows.
- Architected a KV-cache-aware request router that tracked cache locality across distributed nodes, raising simulated cache-hit rate by 8% and cutting redundant computation.
Systems Engineer Intern
May 2026 – Aug 2026
Amazon · Prime Air— Seattle, WA
- Built signal-analysis algorithms for telemetry from 15+ drone and ground systems, turning operational signals into actionable information engineers could use to diagnose and improve the systems they were building.
- Developed a full-stack, self-service telemetry platform that unified data from 15+ drone and ground systems, giving engineers live visibility into fleet health and navigation state.
- Designed data schemas and metric models, including cost-per-package, to standardize measurement across teams and reduce manual analysis during release cycles.
Software Engineering Intern
Jun 2025 – Aug 2025
Capital One— Dallas, TX
- Built an automation framework that parsed Infrastructure-as-Code repositories via the GitHub GraphQL API, surfacing cross-team cloud service ownership and dependencies at scale.
- Designed a recursive parser that mapped 10+ AWS service relationships into a containerized DynamoDB (NoSQL) pipeline, consolidating scattered infrastructure metadata.
- Shipped a full-stack Vue.js dashboard and companion CLI for resource monitoring, adopted by 50+ engineers and cutting incident-investigation time by 2+ hours per case.
Software Engineering Intern
May 2024 – Aug 2024
Cylus · Railway Cybersecurity— Remote
- Built a real-time packet-capture (pcap) streaming tool in Elixir and Phoenix, cutting R&D analysis turnaround time by 50%.
- Reduced annual AWS spend by $3K through PostgreSQL index tuning, AWS Lambda refactoring, and S3 storage optimization.
- Engineered a multithreaded Driver-Machine Interface (DMI) server, demoed live to prospective enterprise clients at InnoTrans 2024.

Selected work

Projects

Distributed Key-Value Store (Raft Consensus)

Implemented the Raft consensus protocol across a multi-node cluster — leader election, log replication, and heartbeat-based failure detection to keep replicas in sync. Added log compaction and snapshotting to bound memory, with a client API for linearizable reads and writes.

PythonDockerKubernetesAWSGitLab CI/CD

Optimizing Compiler

Built an SSA-based optimizing compiler implementing liveness analysis, dead-code elimination, constant folding, and loop-invariant code motion. Designed a register allocator using interference graphs to handle variable spilling.

C++LLVMCompiler Optimizations

mnk-player

An intelligent Monte Carlo Tree Search variant for m,n,k-games (Tic-Tac-Toe, Connect Four). Used LightGBM models trained on Kaggle datasets to guide strategy selection, evaluating board states and pruning suboptimal moves.

PythonMCTSLightGBMGame AI

Simple shell

A custom Unix shell with advanced command parsing via Lex and Yacc. Supports file redirection, pipes, signal handling, environment-variable expansion, built-ins, line editing, history, and wildcarding.

CLex/YaccSystems

What I work with