Search and AI · Senior SWE · 2024 — Present

• Prototyped agentic, multi-turn experiences at Slack on a nimble team, defining core interaction patterns through rapid iteration and A/B-driven engineering decisions; served as DRI for engagement metrics, quality signals, and productionizing prototype code, generating key insights that directly informed the evolution of Slackbot.
• Successfully led release of first two AI features at Slack to stream LLM responses, setting the standard of streaming across our AI features.

Messaging · Senior SWE 2021–2024 · SWE 2019–2021

• Led development of, implemented, and shipped restricting broadcast mentions (@here/@channel) in messages, the #1 product gap for Slack, requested across 37 enterprise customers. Defined and drove project milestones and deliverables, coordinated alignment across product, design, QA, and 5 engineers, and wrote 80% of backend code. Feature is enabled by customers across 25K teams and 73K Slack channels. Built using Hack and SQL, with Logstash, Prometheus, and Grafana for monitoring and visualization.
• Identified the development of backfills at Slack as a technical overhead problem, and got engineering buy-in to change to a better development process, then built and evangelized the solution. Implemented redesign with generics that optimize for database sharding performance (wrote 100% of code). Reduced duplication in code per backfill by 80% and decreased time to implement a backfill by 75%. Framework adopted by 35+ teams and 65+ engineers at the company, 10+ production table iterators containing over 415TB of data and 40+ use case specific backfills written to date.
• Pitched long standing customer requests and got buy-in from 4 different agile teams to lead and deliver on features. Defined deliverables, spearheaded collaboration, built features, and rolled out 4 feature enhancements to production within 8 weeks, 2 weeks each (wrote 100% of backend code). Features shipped contributed to renewal of IBM contract, granted patent US11599235B1, and received Don't Make Me Think and This Small Thing, Big Win team awards. Features have combined traffic of 550K+ MAU.
• Authorized, as part of 1% of engineers at Slack, to review and approve all production table creation and update requests on Vitess (across 15+ keyspaces storing 1PB+ of data). In-depth understanding of database sharding, load performance and optimization, and distributed systems required.
• Designed, implemented, and promoted mentions inspection troubleshooting tool, leveraged by CE and engineers, to identify notification regressions and debug notification related customer escalations. Improved debugging efficiency by 100%, previously not possible to backtrack mentions generated per message for users.
• Refactored (as part of 3-person engineering team) the processing of mentions at Slack, which handles notification processing for 450+ million messages sent per day.
• On-call for Slack's core product. Responsibilities include investigating and resolving high visibility incidents, triaging and fixing critical customer escalations, and addressing high priority bugs. Onboarded 3 on-call engineers and led team debugging sessions to spread best practices and resources across the team.

• Collaborated with cross-functional team of customer success, designers, product managers, and engineers to design, architect, and build customizable dashboards for hospital operations to display real-time data and predictive analytics.
• Increased accessibility to production data for anyone in the company and our customers by developing a tool that allowed querying of production data with an extensible UI, thereby removing data dependency on engineers and expediting customer data requests.
• Researched, designed, and implemented process to anonymize hospital data for internal and external use and added additional features to time shift data and make it more reusable for the future, thus allowing QA to mimic our production environment for testing, sales to have mock real-time data for customer demos, and engineers to have realistic data to build and test features on.
• Implemented query timeout functionality preventing expensive queries from consuming database resources and causing havoc.

• Designed and developed a reliable ETL data pipeline on the AWS platform to create a data-driven tool to identify trends in global news and provide insight into global events using data from The GDELT Project.
• Built distributed system by connecting multiple technologies to pull raw data from S3 datastore, clean and process data in Spark, store data in Cassandra cluster, and extract data from Cassandra using backend web API.