Projects

Data platform, backend infrastructure, and production AI case studies.

Tech filter

4 project(s) matched.

Lydia Agentic AI Assistant and Event Delivery Platform

Software Engineer · 2025

Built Lydia, an agentic AI assistant for fleet-telemetry analytics, and designed the durable event-delivery platform behind its multi-turn tool-calling workflows.

Delivery control path

Receive100%
Outbox92%
Deliver88%
Replay95%
TypeScriptMySQLClickHouseKafka/MSKRedis StreamsSocket.ioDuckDBEC2OpenTelemetry
Read more →
Tool-calling SQL · ClickHouse · Outbox + MSK

Driver Efficiency Scoring with Double Machine Learning

Software Engineer · Dec 2024 - Present

Built a causal evaluation framework that separated driver behavior from route and weather confounders, producing fair rankings and actionable coaching insights for fleet operations.

Driver ranking fairness

Raw gap
18%
Adjusted
12%
Target
10%
PythonPyTorchScikit-LearnSHAPAWS S3Statistical Inference
Read more →
35+ · $0.14 kWh/km (12%) · $38K+/fleet annually

Event-Driven Telemetry Fan-Out for Fleet Operations

Software Engineer · Dec 2024 - Present

Built Kafka and Redis Pub/Sub fan-out paths for live fleet telemetry, enabling sub-second operator dashboard updates and horizontally scalable real-time processing.

Telemetry fan-out path

Ingest100%
Fan out84%
Serve82%
Tune72%
TypeScriptNestJSRedisKafkaAWS ECSMySQLMQTT
Read more →
1M+/day · Sub-second · 42%

Self-Hosted LLM Inference for Lydia

Software Engineer · 2024 - 2025

Deployed vLLM and llama.cpp on on-prem GPU infrastructure behind EC2, NGINX, and Route 53, replacing hosted inference and reducing Lydia serving cost.

Adoption milestones

Engineers
Queries/day
Cost saved
vLLMllama.cppNGINXRoute 53EC2PythonTypeScript
Read more →
7 engineers · 400+ · $2K/month