Live render · Blueprint template
Rendering modern template…
Rendered server-side · same engine used for PDF export
Bullet-point anatomy — what makes each one land
Why these bullets work
Every bullet follows Action → Metric → Outcome and signals one clear strength to both ATS and recruiters.
Senior Software Engineer
Jan 2023 – Present
Google DeepMind
- RAG search over 5M+ docs (LangChain + Pinecone) — research time ↓35%, saving 800 eng-hrs/quarter.🤖 AI / LLM
- Chain-of-thought prompt framework across 12 services — latency ↓22%, hallucinations ↓61%.⚡ Scale & perf
- Monolith → event-driven (Kafka + Go): 200 → 3,400 req/sec (17×), zero downtime.⚡ Scale & perf
- Claude multi-agent code reviewer: 47% more critical bugs caught, prod incidents ↓29%.🤖 AI / LLM
- Fine-tuned Mistral-7B via LoRA at 85% accuracy — replaced $120K/year tool licence.💰 Cost saving
- Mentored 5 engineers; 40-page LLM wiki adopted by 3 teams across 12 projects.👥 Leadership
Software Engineer II
Jun 2020 – Dec 2022
Microsoft Azure — Cognitive Search
- HNSW vector search in Azure Cognitive Search — 1,200+ enterprise customers in 6 months, drove 18% YoY revenue growth.📈 Business impact
- LLM anomaly-detection cost optimiser — $180K/year saved, manual FinOps effort ↓70%.💰 Cost saving
- Serverless cold-start latency ↓60% via ML-predicted container pre-warming.🚀 Velocity
- 8 REST → gRPC: payload ↓40%, P99 latency 180 ms → 42 ms at 50K RPM.⚡ Scale & perf
- 6 ADRs adopted org-wide; standards used by 120+ engineers across Azure Data.👥 Leadership
Software Engineer
Aug 2018 – May 2020
Zepto (YC W21)
- Prophet + XGBoost inventory prediction — stockouts ↓28%, NPS +12 points.📈 Business impact
- A/B testing framework: setup 2 days → 2 hours, enabling 5× more concurrent experiments.🚀 Velocity
- PostgreSQL query tuning: latency 850 ms → 45 ms (19×) for 50K+ daily users.⚡ Scale & perf
- OpenAI Embeddings product search: nDCG@10 0.61 → 0.84, add-to-cart rate +9%.🤖 AI / LLM
Featured Projects
ZustResume — AI Resume Builder
Go · Next.js · DynamoDB · OpenAI · Claude
- GPT-4 bullet suggestions + Claude ATS scoring at 92% recruiter-agreement accuracy.🤖 AI / LLM
- 1,400+ GitHub stars, 300+ WAU within 3 months of launch.📈 Business impact
LLM Code Reviewer — GitHub App
Python · FastAPI · LangChain · Claude API
- Multi-agent Planner→Reviewer→Critic: 87% precision on CVE-class bugs.🤖 AI / LLM
- PR review turnaround: 6 hours → <10 minutes on 400+ active repos.🚀 Velocity
Recruiter-tested power verbs
Every bullet starts with a strong action verb in past tense (except current role). These are tracked by ATS keyword filters and signal ownership.
Build your version in minutes
Open the builder pre-loaded with this sample data, swap in your own details, and let AI polish every bullet.