Work Experience

Massachusetts Institute of Technology - Software Engineer

Location: Boston, MA
Duration: Febuary 2023 - Present

TL;DR

When chemists start with a desired molecule and work backward to figure out the best pathway forward—is a time and resource consuming task, relying heavily on manual analysis and trial and error. AI and machine learning capabilities can streamline and optimize this process.

Please visit askcos.mit.edu to explore the tool! And askcos-docs.mit.edu to learn more.

• Working under the guidance of MIT Prof. Connor W. Coley’s (Forbes 30 Under 30 Healthcare) research lab and collaborated with the MIT Corporate Consortium (10+ Companies) for Machine Learning for Pharmaceutical Discovery and Synthesis (MLPDS) to develop innovative AI-assisted tools and algorithms that are used by multiple pharmaceutical companies.
• Lead the complete re-design of frontend using Nuxt.js from scratch with many other utilities and helped to make them open source. Developed complete CI/CD pipeline on Gitlab, enabling automated deployments and testing of the tool.
• Employed advanced optimization techniques leveraging PyTorch libraries to fine-tune model hyperparameters and implement sophisticated data preprocessing methods. The systematic optimization efforts resulted in a substantial enhancement of model accuracy and a noticeable reduction in training times.
• Containerization helped to increase scalability by creating multiple replicas of the application and ensured load-balancing by using hybrid algorithm. Converted time consuming endpoints to asynchronous using celery workers and RabbitMQ.
• Added platform observability tools such as cAdivsor, Prometheus, and Grafana to watch container metrics at regular intervals and alert on PagerDuty for any anomalies. Helping get better insights on P99 latencies and ways to reduce them.

Team Photo

Meta - Software Engineer Intern

Location: Menlo Park, CA
Duration: May 2022 - August 2022

TL;DR

Web app development, Distributed systems, Stream processing, Test automation and Logging.

• Creation of Policy Management Tool to manage User’s Special Category Data (SCD), which helped in filtering out certain user’s data based on their district such as GDPR for EU.
• Logging if a particular user impression is sensitive or not for a particular policy by collecting data from all the Facebook apps and thus helping the team to roll out features ahead of time.
• Implementing features such as ability to mutate policy’s information and status along with keeping track of timestamps and engineer’s information making the change. Visualizing those changes live using Gantt Chart.
• Reduced manual effort by creating continuous integration/deployment framework to validate configuration file changes upon mutation. Also, unit test and end to end test both in frontend (React) and backend (PHP).

Team Photo