🗓️ Monthly Review - 2025
January - 2025
- Implemented logging
response time
for all the API calls made to other services to monitorlatency
. - Created new
DynamoDB
table for storing user support details andmigrated
block counts. - Integrated new
HMAC
(Hash-based Message Authenticated Code) API. - Deployed
Meta's
Large Language Model,Llama 3.2
, onAWS Bedrock
and builtAI
writing assistant for agents.
February - 2025
- Designed and built
independenlty
DAZN's firstLarge Language Model(LLM)
application, leveraginggenerative AI
withRetrieval Augmented Generation(RAG)
usingVector Search
, that providesreal time analysis
of customer support queries, therefore helping operations team derive strategies toresolve them quickly
. - Built Yoga Pose Recommender: AI-Powered Vector Search for Yoga Poses, an
AI-powered semantic search
system that helps users explore yoga poses. By leveragingGemini (LLM)
to generate rich, context-aware descriptions and create high-dimensionalvector embeddings
, the project transforms an existingHugging Face
dataset into a deeply intelligent and intuitive search experience. With LangChain JS orchestrating data processing and Firestore serving as a high-performancevector database
, users can retrieve the most relevant yoga poses throughnatural language queries
. Additionally, Google Cloud’s Text-to-Speech technology enhances accessibility by providing immersive audio descriptions. Built withNode.js
, this system seamlessly integratesAI-driven search
,vector embeddings
, and real-time query processing to redefine yoga exploration.