Headline

Senior Software Engineer | Node.js · TypeScript · AWS · Serverless | Backend & Platform APIs | Microservices, Event-Driven Architecture | GenAI & LLM Integration | Vector Search | Search Infrastructure @ 17M+ queries/mo

About

Senior Software Engineer at National Pen (Cimpress), owning the backend platforms behind pens.com — search infrastructure serving 17M+ queries/month and an internal asset API at ~78K requests/day. I also build the LLM and vision-model integrations that sit on top of them. What I've built: - V2 Product Asset Management platform — system of record for every product asset on the site. Cloudinary as asset store, a structured metadata layer as the query interface, IAM-scoped writes. My team owns it; every other product team consumes it. - Algolia search architecture and the PLP migration — 529 indices over 622K records. Earlier cut our Algolia API consumption 50% on weekdays, 70% on weekends by fixing an unthrottled typeahead, batching per-keystroke lookups, and removing a redundant facet call. - An LLM-based image-retouching pipeline, built solo and running in production against brand-consistency rules. - BRAIN — an LLM-orchestrated runtime strategy engine with Qdrant retrieval, using LLMs as a decision layer. I designed it and took the architecture through Director of Engineering, VP, and Director of Product approval. - A vision-language order-intake service (30 min → 2 min per order), and a Pinecone RAG support chatbot later deployed by our chatbot team. - Marketing Cards Platform on Magnolia CMS — slot-aware campaign cards injected into the product grid, with scheduling and per-device targeting. ~1,000 campaigns, no engineer required. Backend: Node.js · TypeScript · Java · AWS (Lambda, API Gateway, SQS, EventBridge, DynamoDB, ElastiCache/Redis, S3, CloudFront, CloudWatch) · MySQL · Redis · Docker · Microservices · Event-Driven Architecture · Serverless · REST API Design · System Design · Caching · CI/CD AI & Search: LLM Integration · Generative AI · RAG Pipelines · Vector Search (Pinecone, Qdrant) · Vision-Language Models · Prompt Engineering · Embeddings · Algolia

Location

Greater Bengaluru Area

AI stack

antigravity

claude code

claude

cursor

notebooklm

openai

Work experience

Senior Software Engineer at Cimpress India, Jul 2023 - Present · 3 yrs 2 mos, Own the V2 Product Asset Management platform (PAM) — the system of record for product assets across pens.com, serving ~78K requests/day to product detail pages, the Algolia PLP sync pipeline, and the recommendations sync pipeline. Own the Algolia search integration — 17M+ search requests/month across 529 indices over 622K product records, including index architecture and the batch sync pipelines that keep them current. Designed and shipped the Marketing Cards Platform on Magnolia CMS: a slot-aware promotional system injecting three card types into the product grid (single-slot, multi-slot, full-row banner) with campaign scheduling and independent desktop/mobile content and visibility rules. ~1,000 campaigns configured to date with no engineering involvement. Awarded Cimpress "High Ownership Award". Shipped an AI image-retouching pipeline running in production, processing product imagery against brand-consistency rules during new-product onboarding. Designed "BRAIN", an LLM-orchestrated runtime strategy engine with Qdrant-backed retrieval — architecture approved through Director of Engineering, VP, and Director of Product Management. Stepped in as acting Technical Lead, mentoring engineers and guiding the delivery of scalable e-commerce solutions.

Software Engineer at Cimpress India, Aug 2021 - Jun 2023 · 1 yr 11 mos, Cut Algolia API consumption 50% on weekdays (700K → 320K daily) and 70% on weekends (400K → 120K), bringing the platform back inside its 125M/year contracted quota. Three levers: debounced a typeahead firing on every keystroke from the third character; collapsed two per-keystroke index lookups into one batched multipleQueries call; removed a searchForFacets round-trip that was redundant under a mandatory filter. Built a server-side API layer that moved third-party credentials off the client, eliminating browser-side key exposure. Improved mobile Core Web Vitals: 41% DOM size reduction and 33% Total Blocking Time reduction by deferring third-party script loading to first user interaction. Engineered a Retrieval-Augmented Generation (RAG) chatbot to deliver accurate, context-aware customer support.

DSC core committee member at Developers Student Club RNSIT, Aug 2020 - Jul 2021 · 1 yr, Core committee member for the Google Developers Student Club at RNSIT, organising technical events and workshops for the student community. Presented as a speaker at our web development workshop.

Founder & Lead - BigO at Self-Employed, Aug 2020 - Jun 2021 · 11 mos, Founded and led BigO, a student coding community at RNSIT focused on data structures and algorithms. Ran regular coding competitions and peer problem-solving sessions, growing it into an active community across batches.

Intern at JPMorgan Chase & Co., Jun 2020 - Jun 2020 · 1 mo

Education

BE - Bachelor of Engineering, Computer Science, RNS Institute of Technology - India

Public links

medium: https://medium.com/@ankitagarwal702

leetcode: https://leetcode.com/agarwal-ankit007