• Engineered a custom low-latency Web Acquisition Platform (Svelte, Python) for Clarity Auto Care; improved lead-to-call conversion by 5x (2% → 10%) with an estimated ~$300k in additional annualized revenue.

• Built an AI-Enhanced Vehicle Inspection Platform integrating LLMs to analyze vehicle photos; reduced average inspection time by 65% (45 min → 15 min), reclaiming ~$36k+ annually per technician.

• Architected an autonomous Multi-Stage Investment Agent (Python, ReAct) that analyzes pitch decks and verifies market data; compressed 60 minutes of human analysis into less 5 minutes at 1/50th the operational cost.

• Built a resilient event-driven data ingestion pipeline on GCP Pub/Sub with exponential backoff and DLQ; achieved 99.9% data availability for a $22M VC fund.

• Launched a rental car analytics platform for Turo hosts; achieved $1k MRR within 2 months by automating market demand analysis (Python), before executing a strategic shutdown due to identified long-term viability risks.

• Integrated a Learn-to-Rank (LTR) model into the search stack and negotiated a Latency vs. Revenue trade-off with product leadership, driving an 8% increase in Add-to-Cart rate (~$10M annualized revenue)

• Initiated & Led the Search Interface Unification, consolidating 3 fragmented codebases into a single Search Gateway service; convinced leadership to prioritize refactoring over features, resulting in a 3x increase in engineering velocity for future iterations.

• Architected migration from ElasticSearch to in-house Lucene engine, eliminating P99 latency spikes (400ms → 200ms), uncapping recall limits for 100% of the corpus; evangelized platform adoption across Ads, Consumer, and Substitutions teams

• Owned the end-to-end Grocery Inventory Ingestion pipeline processing 1B+ unique SKUs daily; tuned Kubernetes batch jobs for high-throughput processing, ensuring strict data freshness SLAs for major retailers (Safeway, Walgreens).

• Mentored 2 interns to full-time offers and successfully advocated for an up-leveled Middle Engineer offer for one report by assigning high-impact critical path projects and defending their performance during calibration.

• Scaled backend team from 1 to 6 engineers. Led engineering team to deliver real-time fall detection system for elderly care facilities; platform reliability and accuracy were key factors in closing $4M Series A round

• Drove cross-functional alignment between Product, Backend, and ML teams, translating business requirements into technical roadmaps.

• Established a Privacy by Design security architecture with signed one-time video access links; initiated and managed a third-party security audit that confirmed zero critical vulnerabilities, transforming security into a key sales asset for enterprise partners.

• Standardized engineering processes by introducing Scrum/Agile methodologies, replacing ad-hoc development with structured sprints and retrospectives to improve delivery predictability.

• Refactored a monolithic Computer Vision pipeline into a multi-process architecture using Shared Memory to bypass Python GIL, enabling stable 24/7 real-time fall detection which secured the company's Seed Round.

• Designed a Reverse RPC distributed architecture for edge devices, enabling reliable remote control of cameras in homes with weak LTE connections and restrictive NAT/Firewall environments.