Event-Driven Emergency Response
RescueFlow
0 backlog · 0 dead letters
A synthetic emergency-response workflow built to make distributed-system failure modes explicit and testable.
- Go
- REST
- WebSockets
Aarsh Sheth · Software engineer · San Jose
I’m Aarsh, a software engineer with 3.5+ years in production. My work sits where reliable backend engineering, observable cloud infrastructure, and applied AI meet.
01
Point of view
The useful part starts after the happy path.
I like systems that explain themselves: APIs with clear boundaries, deployment paths that are repeatable, and enough observability to turn production questions into evidence.
Applied AI belongs in that same engineering discipline. Retrieval quality, response latency, persistence, and monitoring are not supporting details — they shape whether an AI product is useful.
02
Project journal
Four engineering case studies
Backend, cloud, applied AI, and developer tooling — documented as working systems.
Project index04 entries
Chatbot-as-a-Service
ChatCraft
90%+ context accuracy
A retrieval-grounded LLM application for enterprise knowledge access.
- FastAPI
- LangChain
- OpenAI
Smart Security Patrol
SeeCure
92%+ detection accuracy
A distributed security-monitoring platform with simulated patrols and real-time vision analysis.
- Django REST
- React
- YOLOv8
Rust CLI + TUI Search
Semantix Lite
CLI + TUI, one engine
A cross-platform local search tool with recursive discovery and an interactive terminal interface.
- Rust
- clap
- walkdir
More engineering work
Agent systems, rapid builds, and platform experiments.
The archive gathers every documented build while keeping this page focused on four complete case studies.
See all projects03
Professional experience · 2021—2024
Work experience
Three production roles across backend systems, cloud infrastructure, and full-stack delivery.
Full Stack Engineer
Avive Solutions Inc.
Modernize a five-year-old, map-intensive product while improving production reliability.
- Migrated core geospatial services to PostgreSQL and PostGIS, improving map feature accuracy by 60%.
- Reduced Docker image size by 90%+ and deployment time by 40% with multi-stage builds.
- Established centralized ELK observability across 10M+ logs and cut API latency by over 50% by resolving N+1 queries and duplicate calls.
- Deployed containerized services on AWS EC2 with CloudWatch monitoring, maintaining 99.9% uptime.
- Refactored a 5+ year-old Django backend around REST and Django REST Framework best practices.
Backend Engineer
Zartek Technologies Pvt. Ltd.
Connect customer-facing products to authentication, finance, document, and subscription services reliably.
- Developed production Django REST APIs for bookkeeping, document-processing, and subscription workflows.
- Integrated Veryfi, QuickBooks, RevenueCat, Firebase Authentication, and FCM.
- Reduced API latency by 35% with AWS Lambda, S3, and Redis-backed serverless workflows.
- Added Pytest coverage and documented APIs with Swagger and OpenAPI.
Backend Engineer (Python)
Fero.AI
Support logistics operations with secure APIs, asynchronous work, and route-aware data.
- Engineered Django REST services with authentication, RBAC, asynchronous APIs, and Redis.
- Modeled PostgreSQL and PostGIS routing data and developed Vue.js delivery-tracking interfaces against the Django REST backend.
- Improved API performance through query optimization and asynchronous processing.
04
Technical toolkit
Technologies I work with
- 01
Backend systems
Production APIs, asynchronous work, and testing
Built and maintained production Django services across three engineering roles.
- Python
- Django
- Django REST Framework
- FastAPI
- Celery
- REST APIs
- GraphQL
- Pytest
- 02
AI & agentic systems
Retrieval, orchestration, and applied models
Built retrieval, vision, multi-agent, and autonomous-agent systems.
- Agentic Workflows
- Multi-Agent Orchestration
- CrewAI
- RAG
- Retrieval Pipelines
- Vector Search
- LLM Integration
- LangChain
- QLoRA
- YOLOv8
- Whisper
- 03
Cloud & reliability
AWS delivery, containers, and observability
Deployed and monitored workloads across EC2, Lambda, RDS, S3, and CloudWatch.
- AWS EC2
- AWS RDS
- AWS S3
- AWS Lambda
- CloudWatch
- Docker
- Kubernetes
- GitHub Actions
- ELK
- Prometheus
- Grafana
- Ansible
- 04
Databases & retrieval
Relational, geospatial, cache, search, and vector storage
Designed production relational, geospatial, cache, search, and vector-retrieval paths.
- PostgreSQL
- PostGIS
- Redis
- MongoDB
- Elasticsearch
- Pinecone
- ChromaDB
- SQLite Vector Storage
- 05
Languages & interfaces
Product interfaces and developer tooling
Shipped web interfaces and local developer tools across TypeScript, JavaScript, and Rust.
- TypeScript
- JavaScript
- Rust
- React
- Next.js
- Angular
- Vue.js
- 06
AI development tools
Agent-assisted implementation, debugging, refactoring, and review
Used daily while reviewing generated changes, validating correctness, and iterating on solutions.
- Claude Code
- OpenAI Codex
- Antigravity
- Devin
- Cursor
05
Beyond the day job
Hackathons & achievements
Fast builds, community contributions, and the work beyond production roles.
Selected record
Awards & contributions
- Winner
DATA ER — 1st Place Overall & Hotdata Track Winner
Data & AI Hackathon, AWS Builder Loft San Francisco
Built a multi-agent data repair system with RocketRide orchestrating parallel specialists and Hotdata providing isolated per-agent databases and persistent cross-run telemetry for cleaned, auditable datasets.
- June 2026
Viralyst — 2nd Place
InsForge Hackathon, Entrepreneurs First
Led an eight-hour build combining Whisper, vision models, Postgres, and vector similarity search to simulate video virality.
- May 2026
SMOKY — 2nd Place
Applied Intelligence Hackathon
Led a ten-hour build of an autonomous DevSecOps agent using Opsera MCP, Daytona, NEAR AI, React, FastAPI, and InsForge.
- April 2025
Legal QA Engine — Top Ranking
LLM x Law Hackathon #5, Stanford University
Led a four-person team to deliver a GPT-4 legal QA engine with LangChain and vector search in nine hours.
- Contribution
Faker — Open Source
Python community contribution
Added Hindi, Gujarati, and enhanced English-India locale support with tested unittest modules.
Alongside the work
Education, leadership & service
Education
San José State UniversityMaster of Science in Software Engineering
Graduating December 2026
Leadership & teaching
President of Spartan Garba Club, coordinating volunteers for an event with 500+ attendees. Teaching Associate and ISA for CMPE 126: Data Structures & Algorithms.
Community
iSucceed peer mentor for international students and volunteer event photographer with ISSS.
A personal invitation
Have a hard backend problem or an AI idea that needs engineering discipline?
I’d be glad to compare notes.
contact@aarsheth.com