Skills

Applied AI: LangGraph, LiteLLM, Langfuse, MCP, Anthropic API, Claude Code, Pi

ML & Serving: vLLM, FastAPI, MLflow, Kubeflow Pipelines, DVC

Languages & Backend: Go, Python, TypeScript, SQL, Bash, Node.js, Bun, React, Next.js, Vue.js, Redis, Kafka

Platform & Cloud: Kubernetes, ArgoCD, Crossplane, Helm, Terraform, Temporal, KEDA, Docker, AWS, GCP, Azure, GitHub Actions

Observability & Data: Prometheus, Grafana, OpenTelemetry, ClickHouse, Postgres

Professional Experience
Software Engineer II
Mar 2026 – Present | Bangalore
  • •Migrated an internal agent platform from closed frontier models to open-weight models (DeepSeek V4.1 Flash and GLM-5.3), cutting token spend by ~$70k/month with no measurable drop in task success rate on internal evals.
  • •Designed a deep agent for early detection and root-cause analysis of anomalous cloud cost spikes, saving ~$20k/month.
  • •Built an AI agent for long-running infrastructure migrations, saving 50% of engineering time. The agent chooses deployment rings, tracks progress, and needs minimal human intervention across customer maintenance windows in different timezones.
  • •Architected Release Management tooling to track per-tenant deployment state for 700+ enterprise customers, eliminating on average ~5 Sev0 incidents per quarter caused by unverified upstream rollouts and unclear ring deployment status.
  • •Scaled Argo CD to 1000+ applications across EKS, GKE, and AKS; cut sync time by 20 minutes/app through fine-tuning and an Argo CD upgrade (v3.1.2 → v3.2.6).
    • Software Engineer I
    Jun 2025 – Feb 2026 | Bangalore
  • •Owned development and reviews for Atlan's Internal Developer Portal (IDP) using Vue.js and Go, serving 1000+ users across EPD, Sales, and CX; established IDP as the source of truth for tenant metadata.
  • •Achieved 90%+ success rate per month in on-demand infrastructure provisioning workflows (team KPI for 6+ months), up from a baseline of ~50%, by fixing Terraform modules and making bootstrap jobs idempotent for the initial setup.
  • •Moved the infrastructure provisioning flow to Crossplane to further achieve a 100% success rate.
  • •Extended the AI support chatbot to automate 73% of weekly support tickets by auto-generating PrivateLink setups and surfacing tenant details such as AWS ARNs and NAT IPs.
    • Associate Software Engineer I
    Nov 2024 – May 2025 | Mumbai
  • •Integrated OpenTelemetry (OTEL) agent shipping logs to ClickHouse for IDP.
  • •Wrote a Temporal workflow to monitor resource quota across clouds with Prometheus metrics + Grafana alerting, catching 100% of provisioning or scale-up failures before exhaustion.
    • simPPL⁠, Frontend Intern
    May 2023 – Aug 2023 | Remote, USA
  • •Created data-rich visualisations (1 million+ rows) using D3.js for misinformation datasets (X, Facebook, Telegram).
  • •Built discord-triggered content pipeline: web-scraped articles, summarised via LLM & distributed as newsletters to 300+ subscribers.
    • Projects
  • •Autonomous research loop (inspired by Karpathy's Autoresearch) where Claude proposes changes in the model and promotes candidates that beat the current champion – improved model accuracy from ~70% → 80% across 20 iterations.
  • •Uses DVC for data versioning, Kubeflow Pipelines for training, and MLflow model registry (champion/challenger).
  • •FastAPI inference serving via GitOps that auto-rolls new champions on MLflow promotions; zero human in the loop.
  • •Go modular monolith with domain-separated services (booking, matching, pricing), using Kafka for event-driven communication, async booking processing, and real-time ride tracking with Redis as a distributed lock for rider bookings.
  • •Leveraged the Google S2 library for efficient spatial indexing and proximity searches in the driver-matching service.
    • Achievements
    Education
    Dwarkadas J Sanghvi College of Engineering, B.Tech in Computer Engineering
    Sep 2021 – May 2025 | GPA - 8.13 / 10
    HSC, JEE Main and Advanced
    2020 – 2021 | Self-taught with AIR of ~18,000