Senior Software Engineer
Send a job offer directly to this candidate
Senior Software Engineer with 13+ years of experience shipping B2B SaaS products and API‑first platforms at Perplexity (AI search), Plaid (fintech), and Okta (identity). Deep expertise in LLM systems, GenAI infrastructure, API design, and product engineering for building scalable systems that other engineers and customers rely on. Proven ability to own end‑to‑end features, lead cross‑functional efforts, and deliver at startup speed.
Seeking senior IC role to drive system design, mentor engineers, and build impactful product features.
Perplexity AI | Senior Software Engineer
Dec 2022 – Present
● Delivered end‑to‑end product engineering for Sonar API, designing a multi‑model router (20+ endpoints) with circuit breakers and cost‑aware policies, improving TTFT by 2.9× and enabling B2B enterprise customers to integrate in hours.
● Architected system design for RAG retrieval layer, combining embedding‑based and vector search over a 200B+
document index, improving answer grounding and reducing hallucination through citation injection.
● Enhanced LLM serving for cost & scalability, implementing disaggregated prefill and decode along with FP8
quantization (Triton + TensorRT-LLM), reducing per‑query cost by about 5× and improving SaaS unit economics.
● Launched developer console features, including API key management, usage dashboards, and a live API playground with SSE, improving developer onboarding time by 40% and increasing API adoption among B2B customers.
● Led cross‑functional reliability efforts, collaborating with product and ML teams to implement health checks, canary routing, and regression gates, reducing post-incident p99 latency regression by 90% while mentoring 2 engineers.
● Scaled generative AI systems to 200M+ daily requests with sub‑second TTFT, maintaining weekly release cadence and cutting feature lead time by about 40%.
Stack: Python, PyTorch, TypeScript, Next.js, Tailwind CSS, Vespa, Triton, TensorRT-LLM, vLLM, Kubernetes, AWS Bedrock
Plaid | Senior Software Engineer
Aug 2017 – Dec 2022
● Orchestrated Kubernetes migration and API design, building a self‑serve platform (YAML DSL, Argo Rollouts, Linkerd)
and migrating 100+ microservices from ECS, cutting deploy-time to 30 minutes and enabling independent releases.
● Optimized Node.js bank‑integration service, diagnosing connection pool and GC bottlenecks using clinic.js, increasing concurrency 30×, saving about $300K per year in EC2 cost, owning the fix end‑to‑end.
● Authored react-plaid-link SDK, the official React wrapper with TypeScript definitions and usePlaidLink hook, used by thousands of B2B developers and reducing integration friction by 70%.
● Core contributor to Plaid Signal (ACH risk scoring), building real‑time feature pipeline and XGBoost inference API with p99 latency under 50ms, monitoring drift with Prometheus and adding SHAP explanations for FCRA compliance,
helping scale the platform to process $1.5B per month within the first year, becoming a key B2B revenue driver.
● Drove deployment safety culture, automating rollback triggers when error rate exceeded 1% or p99 latency,
exceeded 2× baseline and leading blameless post‑mortems, reducing MTTR by 40%.
● Spearheaded cross‑team RFCs for Kubernetes migration, influencing architecture decisions across 3+ teams and improving microservices boundaries and observability.
Stack: Go, Python, Node.js, TypeScript, React, PostgreSQL, Redis, Kafka, AWS EKS, Terraform, Linkerd, Argo Rollouts
Jul 2012 – Jul 2017
● Owned end-to-end product engineering for Adaptive MFA policy engine, evaluating real-time risk signals and driving
40% YoY MFA adoption while reducing login friction for B2B enterprise customers.
● Shipped an OIDC-certified OAuth 2.0 authorization server with multi-tenant support and 6 grant types, forming
Okta’s API Access Management Product and enabling secure APIs for thousands of enterprise apps.
● Developed Okta Verify Push MFA with device‑bound keys in Secure Enclave/Keystore, challenge‑response JWT and replay protection, reducing MFA authentication time from 15s to 3s, improving UX for B2B SaaS customers.
● Engineered multi‑tenant isolation with per‑tenant rate limiting using a Redis token bucket and request‑ID correlation,
reducing cross‑tenant incidents by 80% while maintaining 99.99% uptime during pre‑IPO scaling.
● Built core frontend product features including Okta Sign‑In Widget deployed across 1K customer login pages supporting 2M+ daily logins, along with admin console modules for SSO, MFA and lifecycle management.
● Collaborated with product and design to deliver admin workflows and authentication UI that improved customer satisfaction.
Stack: Java, Spring, MySQL, Redis, AWS EC2, JavaScript, Ember.js, iOS Swift, Android Java
University of North Carolina, Chapel Hill
B.A. in Computer Science
2012
