recruiting platform infra jazz and having fun with ai agents :) hehe

• Developed a scalable solution to migrate customer data from competitor platforms to our next-generation platform, using Golang and Redis for caching.

• Built a QA solution to verify the success of data migrations across a complex backend architecture, ensuring data was available and functional on our platform.

• Successfully migrated and quality-checked over one million records from our legacy system to the new platform, achieving 99.99% accuracy and data availability.

• Created a deployment bot that integrated our entry-level platform into smaller customers’ infrastructure, leveraging advanced technologies like LangChain, LangGraph, Facebook Llama models, AWS EKS, and NeMo Guardrail with Python. This reduced setup time from two months to just one week.

• Mentored junior interns, providing guidance on system design and developing scalable systems for Kubernetes deployment.

Implemented Large Language Model (LLM) for Log Data Analysis: Deployed the CodeLlama model on Google Colaboratory to analyze log data and extract actionable insights using SQL on the Snowflake database. This involved setting up the model in a collaborative environment, enabling efficient analysis and reporting.

Developed a Semantic Search Space with AWS OpenSearch: Created an embedded vector space using AWS OpenSearch to enhance LLM capabilities. This setup allows the model to query relevant terms based on specific input words, improving the precision and relevance of the responses.

Built an AI-Powered Log Data Analyzer: Utilized AWS SageMaker, AWS OpenSearch, Streamlit, and LangChain to develop an LLM for detecting anomalies and identifying root causes (RCA) in log data. This solution enhances troubleshooting processes by quickly pinpointing critical issues.

Backend development for B2B Partner's platform in Golang

• Collaborated with a variety of skilled individuals ranging from product designers to analytical engineers

• Created a batch processing system using AWS SQS and AWS SNS to decouple multiple AWS Elastic Beanstalk servers/instances at scale.

• Using batch process system aggregated 12K messages to be delivered over various mediums all under 700ms using Golang's concurrency.

• Added a login token management system for the platform to allow users to log in through our respective login provider or their own IdP.

• Allowed said user access privileges based on roles using the least privilege access model.

Diamond J Dairy came to our university with a request to solve their water conservation problem: Leverage automation to conserve water in soaker areas within the cow pens to replace the current manual process.

• Within the water circuit, designed a Python algorithm to open a water valve if it detected livestock in said pen otherwise moves to the next pen and rechecks.

• Used OpenCV algorithm along with You Only Look Once's pre-trained model to detect if a cow is present in a pen.

• Utilized LabelImg's annotation software along with Google's Colaborator's GPU, I trained cow models specific to Diamond J Dairy Farm. Increasing detection rate by 30%.

• Solution was accepted and saved hundreds of gallons of water daily