Oct 2024 – Present · 1 yr 5 mos · Bengaluru, Karnataka, India · On-site

• Leading AI Initiatives for Customer Support Ecosystem. Specialising in RAG bases systems and agentic workflows blending complex SOPs with real-time user data for personalised assistance. Hence, 80% tickets are handled by AI agents now with optimal CSAT score as control.
• Building NLQ Agents to allow support team to query and aggregate real time user data using plain English from our storage solutions to improve ticket resolution speed.
• Led end-to-end resolution of Timesheet system as technical lead, product manager, and UX designer, ensuring accurate shift logging and payment processing for temp workforce
• Created Trust Framework to automate decision-making for high-quality users.

Sep 2023 – Nov 2024 · 1 yr 2 mos

Feb 2023 – Aug 2023 · 6 mos · Bengaluru, Karnataka, India · On-site

• Successfully developed a centralized support dashboard, resulting in a remarkable 20-30% reduction in average handling time across diverse workflows.
• Spearheading the automation of processes via Chat Bots, streamlining operations by eliminating agent intervention or gathering context for swift query resolution.
• Worked on several proof-of-concept projects using ChatGPT, enhancing support interactions and delivering personalized experiences within the support ecosystem.

Jul 2022 – Nov 2022 · 4 mos

• Lead a team of 4 Engineers all focussed on building MVP for "Sherlock" - fully loaded social search across crypto communities to help analysts take bolder & prudent decisions. https://app.getsupermind.com/
• The whole product involved continuously scrapping data from social platforms (like Telegram, Twitter etc), indexing and then eventually serving them using a search interface to power conversational analytics like experience with average latency of ~300 ms.
• Worked with search solutions like Vespa and ElasticSearch to power the search functionality and developed a custom ranking algorithm to rank the messages.
• Leveraged DS, NLP techniques like POS tagging (Spacy) to custom annotate data to improve readability and NLI Models to bucket messages into different buckets.
• Setup Analytics tools like Clarity and Mixpanel to track different KPI's.

Mar 2022 – Jul 2022 · 4 mos · Bengaluru, Karnataka, India

• Part of the initial Product/Tech team, hence worked on building and delivering many foundational constructs regarding the product lifecycle in coordination with other segments like Business, Sales, Customer Experience etc.
• Built a robust Rule Engine from scratch using MVEL Expression Language which powered many promotional constructs like coupons, sale events, referral system etc. This allowed for better flexibility in customising such constructs.
• Switched our Postgres DB from a Single Master to a Master + Hot Standby architecture to improve on reliability and availability. Setup monitoring tools like PgHero to setup necessary indexing and identify/fix hot queries.
• Setup OLAP infra using OSS tools like Clickhouse (Columnar Database optimised for OLAP) and then connected it with tools like Metabase which helped us build powerful business intelligence dashboards that allowed us to find long term trends and track our KPI's to help product decisions.
• Used tools like JHipster to fast-rack our development lifecycle which allowed us to generate boilerplate code based on a schema, allowing us to focus more on developing business logic.

Jul 2019 – Mar 2022 · 2 yrs 8 mos · Greater Bengaluru Area

• Worked alongside Business team and other tech teams to work on many product features like Merchant Catalog Management, Combo Product Feature, Dynamic Product Pricing etc which resulted in considerable movement in product KPI's.
• Conceptualised, Built and Scaled a quality brand-pivoted distribution program "Superclub" from scratch. This involved an all round effort in all parts of the product with constant iterations supported by data and market feedback. This improved the GMV by 65%.
• Worked and built a fully-automated in-house SQL monitoring tool "Dr S" which identified issues in SQL queries during the code review phase and allowed the developer to analyse query performance using customised consoles to avoid issues in production env. This involved working with Detekt (Static code analyser for Kotlin), Snorql (OSS SQL Performance utility by Udaan) and Github API's.
• Developed a in memory solution - Trie Based approach which allowed us to deduce the characteristics of all shipments between any two pincodes, hence allowing us to manipulate price of products at runtime with average lookup time of 0.6 ms, ~200 Mb memory footprint and 3.7 s build times. Also published a blog for this https://medium.com/engineering-udaan/try-trie-and-youll-succeed-24db9b51a6d6
• Developed an in-house Audit Framework based on Myers Diff Algorithm which allowed us to just persist the changes as a general purpose diff and allowed us to recreate the state at runtime from the persisted record in DB. We also persisted necessary telemetry and audit properties to understand the reason and change initiated. This method also allowed reversion out of the box

Jan 2019 – Jun 2019 · 5 mos · India