Shashank Yadav
Job Titles: Senior Software Engineer, Senior Backend Engineer, Backend Engineer, Ruby on Rails Developer, Rails Engineer, Platform Engineer, SDE II, Software Development Engineer, Full Stack Developer
Core Skills: Ruby, Go, Ruby on Rails, Rails 4, Rails 5, Rails 6, Rails 7, Apache Kafka, kafka-batch, kafka-batch-go, Karafka, Sidekiq, Sidekiq Pro, Redis, MySQL, SQL, read replicas, tiered caching, background jobs, async processing, job queues, batch processing, bulk inserts, concurrency tuning, queue topology, fair scheduling, multi-tenant fairness, idempotency, checkpoint orchestration, pause resume jobs, failover, disaster recovery, capacity planning, query optimization, schema optimization, N+1 queries, API throughput, REST API, multi-tenant, Docker, containerized services, Databricks, data warehouse, MongoDB, 100 million records, 200 million jobs per day, infrastructure cost reduction, pod optimization, Kubernetes, distributed systems, microservices, system design, software architecture, three-tier architecture, monolith modernization, Rails engines, Ruby gems, TDD, test driven development, Agile, Scrum
AI Skills: Cursor Pro, AI-assisted development, agentic AI, AI agents, agentic workflow, MCP Model Context Protocol, Jira integration, Airbrake integration, automated triage, pull request automation, code review automation, productivity engineering, human in the loop, secure AI adoption
Companies: Punchh, PAR Technology, Daffodil Software, CampusBox
Education: B.Tech Bachelor of Technology, Galgotias College of Engineering and Technology, GATE 2018 Qualified
Location: Gurugram, Gurgaon, Haryana, India, NCR
Contact: shashank0x1@gmail.com linkedin.com/in/iamshashankio github.com/iamshashank
Impact Metrics: kafka-batch Sidekiq Pro alternative under 50 percent cost, 35 percent pod reduction, 35 percent infrastructure cost savings, 200M+ background jobs daily, 100M+ row datasets, checkpoint resume, tiered caching, read replica routing
Recognition: Hero Award Punchh PAR 2020 2021
Professional Summary
Senior Software Engineer with 8+ years designing and operating high-throughput
Ruby on Rails backends. Strongest signal: system design & architecture—designed and built
kafka-batch (Ruby/Karafka) and kafka-batch-go, a three-tier Sidekiq Pro–class batch platform on
Kafka + Redis with mixed-runtime execution, fair multi-tenant scheduling, and ~50% lower cost under comparable load.
Deep production experience with Sidekiq at 200M+ jobs/day, tiered caching, read replicas, failover, and
checkpointed bulk pipelines—including ~35% fewer worker pods via batching and utilization work.
Ships with agentic AI workflows (Cursor Pro + MCP) for triage, debugging, and delivery. GATE qualified (2018).
Technical Skills
- Languages
- Ruby, Go, SQL, JavaScript (prior full-stack)
- Frameworks
- Ruby on Rails 4.x–7.x, Rails engines & internal gems, Karafka
- Async & messaging
- Kafka, kafka-batch / kafka-batch-go, Sidekiq (Pro patterns), Redis, fair scheduling, job dedup & locking
- Data & scale
- MySQL (query/schema optimization), read replicas, tiered caching, bulk pipelines, 100M+ row processing
- Architecture
- Three-tier control/execution planes, multi-tenant fairness, idempotent at-least-once design, capacity planning
- Reliability
- Checkpoint orchestration, failover design, workset reclaim, retries & DLQ
- Platform
- Multi-tenant services, Docker, API throughput tuning, TDD, Agile
- AI & tooling
- Cursor Pro, MCP agent orchestration, Jira & Airbrake integrations, AI-assisted triage & code review
Architecture Highlight — kafka-batch
Designed end-to-end a Sidekiq Pro Batches replacement on Kafka: durable job transport, Redis coordination,
and wire-compatible Ruby + Go runtimes that can share a single batch.
- Architected a three-tier model (client / control / execution) so each plane is independently deployable; Ruby and Go communicate only via Kafka topics + Redis contracts.
- Built kafka-batch (Ruby gem on Karafka) and kafka-batch-go (
kbatch daemon & workers)—same job envelope, batch ledger, handler manifest, and completion semantics across languages.
- Designed multi-tenant fair scheduling, priority queues, SuperFetch claim/mark/perform, retries/DLQ, delayed jobs, and Sidekiq-style batch callbacks (
on_success / on_complete) for production safety under at-least-once delivery.
- Targeted under ~50% of Sidekiq Pro TCO at similar load by moving durability to Kafka and right-sizing the control plane—built as a platform others can adopt, not a one-off script.
AI Assistance & Agentic Workflow
Hands-on with AI-augmented development—building practical agent workflows that compound engineering output
while respecting enterprise security boundaries.
- Built custom AI agents on Cursor Pro with MCP integrations to Jira and Airbrake that monitor assigned tickets, correlate production errors with local codebase context, and surface likely root causes before deep manual investigation.
- Designed an end-to-end agentic workflow where agents analyze code on the local machine and produce suggested PR fixes with detailed comments and rationale—compressing time from ticket assignment to review-ready change sets.
- Enforced strict data-access guardrails per company policy: production database access is not granted to local AI agents, prioritizing productivity without exposing sensitive systems.
- Continuously refines agent prompts, tool boundaries, and human-in-the-loop review to improve signal quality, reduce toil, and increase ship velocity across backend and async job systems.
Key Impact
| kafka-batch platform | Designed Ruby + Go Sidekiq Pro–class batching on Kafka at ~50% lower cost |
|---|
| Sidekiq / workers | Higher throughput; ~35% fewer pods for comparable load |
|---|
| Data scale | Background jobs over 100M+ records with checkpoint resume |
|---|
| Platform volume | Systems operating at 200M+ jobs/day throughput |
|---|
| AI productivity | Custom Jira/Airbrake agents for faster triage and PR-ready fixes |
|---|
Professional Experience
Senior Software Engineer (Mar 2024 – Present) · SDE II (Nov 2021 – Present)
- Designed & architected kafka-batch + kafka-batch-go—an in-house Sidekiq Pro replacement on Kafka with Batches, fair multi-tenant scheduling, mixed Ruby/Go execution, and under ~50% the cost at similar load.
- Scaled async processing for a platform running 200M+ background jobs/day by redesigning Sidekiq workflows—queue topology, batching, concurrency tuning, and hot-path refactors.
- Reduced infrastructure cost ~35% by optimizing worker efficiency and right-sizing pod counts while maintaining SLAs—profiling hot paths and adding safe pause/resume checkpoints to long-running jobs.
- Introduced tiered caching and read-replica routing to offload primary database pressure and improve API and job throughput under peak campaign load.
- Revamped internal systems onto Databricks so a single background job can safely process 100M+ row datasets with checkpoint-based orchestration—resume after failure without full restarts.
- Designed failover when Sidekiq Redis memory reached critical levels so critical pipelines stay available under memory pressure.
- Extracted reusable failover and checkpoint components for cross-service adoption.
Stack: Ruby on Rails, Kafka (kafka-batch / kafka-batch-go), Go, Sidekiq, Redis, MySQL (+ replicas), Databricks warehouse, MongoDB, POS, multi-tenant configuration, containerized services
Associate IT — Full Stack Ruby on Rails
- Upgraded Rails monoliths from 4 → 5 → 6; improved stability and long-term maintainability across e-commerce and real-estate products.
- Reduced page load times by fixing N+1 queries, schema refactors, Redis caching, and strategic memoization.
- Modularized monoliths by extracting reusable domains into Rails engines packaged as internal gems for cross-project reuse.
AngularJS Frontend Developer (Intern)
Education
B.Tech — Galgotias College of Engineering & Technology
2014 – 2018
GATE 2018 — Qualified
Recognition
Hero Award — Punchh / PAR (2020–2021)