Riken Shah
Performance Engineer
I build systems that scale to hundreds of thousands of people. Founding Engineer at Tasks, working on distributed systems, local-first architecture, and AI infrastructure.
Testimonials
"Riken is high energy, high passion person. He audited and re-architected the entire infrastructure for Gratitude (8M+ users, 54M+ requests/month, ~4TB egress/month), bringing AWS costs down 8x and making content delivery 2x faster. All in two weeks with zero downtime. He understands systems deeply and delivers results fast."
"Riken helped me reverse engineer and fix a critical system that was breaking a core feature for users. His work restored the extension's ability to find and save offers for CardPointers (1M+ users) with a troublesome bank connector, helping users save $100s every month. It was amazing to see how quickly he diagnosed the problem and we were able to ship a fix together the same day. Highly, highly recommend working with Riken if you have the opportunity!"
"Riken single-handedly built the entire backend for Tasks (700K+ users, 4.8+ stars, featured in 100+ countries) in Golang from scratch and owns all of our infrastructure: deployments, monitoring, architecture, everything. In 3+ years of working together, I've never had to worry about the backend. After his work on the sync engine, user complaints about syncing dropped by 90%. He's the kind of engineer you trust completely with production systems."
Writing
Research
Talks
Recognition
Projects
Ventures
Built an AI-powered value chain for the Indian textile industry. Designed a 100TB image indexing engine for local search. Signed the first client after 9 months and scaled to 9 countries.
Experience
Scaled Tasks from 100K to nearly 1M users on AWS while revenue grew 200%; Apple-featured in 100+ countries.
Built the sync engine from scratch using a modified CRDT architecture in Go. Replaced CloudKit to support real-time collaboration with sub-500ms p99 latency globally and zero data loss incidents, cutting sync complaints by 90%+.
Built the web app with a viewport-based rendering system scaling from 100 to 100K tasks under ~150MB memory. Rewrote the backend from Node.js to Go, achieving a 4.6x latency improvement (p99: 1s → 220ms).
Integrated observability with the LGTM Stack (Loki, Grafana, Tempo), resolving 80% of recurring production errors. Built a production ReAct agent on Amazon Bedrock AgentCore handling 100K queries/day at peak. Built a real-time voice-controlled AI agent using Gemini Live multimodal API with sub-200ms action latency. Built Yagumi, an AI flow orchestration layer with a 52-flow registry and DAG-based plan builder. Owned infrastructure end-to-end, 99.99% uptime.
Architected a distributed web scraping system with 50+ Chrome instances across 20 nodes, processing 1M+ pages/day at 95%+ success rate. Built an automated PlayStore deployment pipeline managing 5,000+ apps, productized as SaaS at $80K/yr ARR.
Optimized file storage via compression and S3 lifecycle policies, cutting costs by 76% ($5K → $1.2K/month). Took ownership of backend architecture after manager departure, leading 10 core services at 99.99% availability.
57+ merged PRs and 100+ commits across full-stack features, CI stability, and major refactors. Built the Emoji Status feature that headlined the Zulip v5 release, now used by 1000+ organizations. Selected for Google Summer of Code at age 17 (2021, 2022), then selected as one of 26 mentors for GSoC 2024.