Amogh Jahagirdar

backend engineer - billing, monetization, platform integrity

Amogh Jahagirdar

01Contact

San Francisco, CA
me [at] amogh [dot] info
GitHub
LinkedIn
Resume (PDF)

02About

I'm a backend engineer at Luma AI, where I work on billing, monetization, and platform integrity. Previously I built payments and crypto infrastructure, including the Android side of an open-source Stripe library for React Native that has processed $1.5M+ in transactions for more than 32k customers. Competitions Expert on Kaggle, with a silver medal in the top 1% of 1,600+ teams for Toxic Text classification.

03Work

Luma AI

Backend Engineer #3, Feb 2026 - present

Billing, monetization, and platform integrity for the agent platform.

  • Backend hire #3; own billing and monetization for Luma, supporting 350k MAU on the Agents product.
  • Built enterprise Pay-As-You-Go billing end to end, enabling the sales team to sign 17 new clients who spent $120k in the first 2 months.
  • Cut p95 usage-period close time 16,000x, from 607 hours to under 2 minutes, by redefining core billing invariants and unwedging a deadlocked settle pipeline.
  • Unlocked $1.5M in overage spend by moving spending-cap enforcement from post-spend reconciliation to real-time checks, fixing overruns that hit 27% of users and had made the feature unusable.
  • Implemented a content moderation pipeline that scans 12.5M+ artifacts and enforces 108k+ policy refusals per month, pairing direct Hive classifier integration with an audit trail that records every decision.
  • Built watermark enforcement for free-tier (6M+ artifacts/month), driving a 22% increase in conversion.
  • Designed and shipped LLM moderation for Luma's Generation API, screening 1.6M+ monthly requests to the Ray and Uni model families before generation.
  • Raised trial activation 3.8x by replacing the credit-card wall with a verified phone gate which cut repeat abusers off from 23M free-trial credits.
  • Shipped Luma's first third-party GPU inference path, serving 1M+ Ray3 video generation jobs per month.
  • Built per-customer LLM cost attribution by logging usage at the inference boundary into a user-tagged schema, capturing 36M records/month which fed data science's COGS tracking.
  • Created a framework that lets product teams add modals, banners and prompts to the Luma app, now used by 20 engineers and serving 400k calls/day.

04Earlier work

Slingshot Finance

Backend Engineer (acquired by Magic Eden)

  • Early engineer #10, serving 1M transacting MAUs at its peak.
  • Designed and operated a distributed, event-driven backend with 100+ deployed services on Kubernetes; the core trade engine service supporting 10k RPS.
  • Built and maintained blockchain data pipelines which ingested data from 10k+ sources concurrently using Kafka, Redis and Postgres.
  • Engineered gas-aware routing algorithms, improving user net profits by up to 15% through optimal multi-hop pathing and dynamic gas estimation.
  • Scaled a high throughput rewards platform 20x (from 700 blockchain txs/min to 15k/min), powering multiple promos serving 80k+ users, and driving 50x growth in volume, $20M+ in increased volume.
  • Tech lead for multi-chain expansion, leading a team of 2 developers, and driving cross-functional launches integrating 20+ blockchains to maximize product surface.

Magic Eden

Backend Engineer

  • The largest NFT marketplace by trading volume ($3.2B in 2024, per CoinGecko).
  • Built a fault-tolerant transaction execution system using NestJS with full observability (distributed tracing, custom metrics) and sub-second recovery from failures.
  • Integrated the new engine into Magic Eden's core infra, scaling to 100k+ on-chain ops/day with queue-based retries, 3x throughput improvement, and 99.99% uptime.

Freelance Developer

  • Built and maintained commercial React Native apps used by 1M+ users.
  • Designed and deployed a custom GraphQL + Stripe SDK backend powering in-app payments.

05Projects

06Education

National Institute of Technology, Trichy, India

Bachelor of Technology

  • Admitted by ranking in the top 0.2% nationwide.
  • Graduated in the First Class with Distinction, with a cumulative GPA of 8.81/10.

07Talks

08Open source

Ethereum developer tooling

Early contributor and maintainer of scaffold-eth, the leading Ethereum developer onboarding framework. Built reference projects that became core learning templates (9K+ GitHub stars and 4k+ forks). Awarded a 1 ETH grant by the Ethereum Foundation.

Packages

09Writing

Multilingual Toxic Comment Classification

Kaggle competition: train on English toxic comments, detect toxicity across languages.

  • Won a silver medal in the top 1% of 1600+ teams, and my Kaggle profile was promoted to Competitions Expert.
  • With little usable multilingual data, we generated our own with translation models.
  • Trained language-specific models on the translated sets and blended the results over many rounds of hyperparameter tuning.