Experience

Part of SecDb team, which is responsible for development and maintenance of the firm's proprietary database, pricing and risk management system.
Implemented Prometheus observability: Defined Prometheus metrics and developed an HTTP server in C++ to expose them. Automated the generation of Prometheus configuration with Go to define scrape targets. Created Grafana dashboards for metrics visualization and configured Prometheus Alertmanager for alerting, enabling the SRE team to effectively monitor the service.
Onboarded structured logs to GCP BigQuery: Implemented FluentBit for log processing and forwarding, defining regex parsers to convert unstructured logs into a structured format. Forwarded logs from on-premises hosts to GCP, and set up BigQuery tables for log storage and analysis using SQL, which significantly improved issue diagnosis.
Automated the synchronization of PagerDuty schedules with an internal calendar, using OAuth functionality to interact with PagerDuty REST API and managing credentials using Credential Vault. Reduced sync time by 90%.
>On-call support: Resolved production issues involving 10,000+ SecDb databases and 2,000+ hosts, including doing database failovers and creating new archive databases. Responded to clients' queries and wrote postmortems.

Designed and implemented, in Scala, an Akka cluster - a decentralized peer-to-peer based cluster membership service which uses the Actor Model and the gossip protocol to meet the demands of modern distributed systems.
Migrated 50+ production jobs to use the newly created Akka cluster for data communication between multiple actor nodes in different servers. Performed regression and performance testing before pushing the code to production.

Proposed BiasHeal, a black-box post-processing method to self-heal bias in sentiment analysis systems. Second author to the paper accepted at ICSME, under the New Ideas and Emerging Result track. Coding in Python.
Worked on BiasFinder, an approach to discover biased predictions in sentiment analysis systems via metamorphic testing and automatic curation of suitable templates using NLP techniques

Developed and deployed, from scratch, a platform to connect law firms with those requiring legal services. Tech Stack: React.js, Redux, Express.js and MongoDB.
Key features include table sorting and filtering, admin approval of law firm reviews, bidding for and adding a listing, and user login and registration.

Hands-on experience with Java and Android Studio to build an application which mimics the classic multiplayer card game Canasta.
Involved in writing backend code for various functions which were part of the game logic.