I'm Charvi, an AI/ML Engineer and Researcher.
Here's a quick intro about me and what I love to do
Finding My Way into AI/ML
The fact that a bunch of tweaked numbers could learn patterns and make smart decisions absolutely fascinated me. I started where most people do, playing around with ML models, CNNs, RNNs, and building things like recommender systems. I compounded my knowledge with research papers, a few patents, and lots of side projects. But it didn't stop there. LLMs blew up AND so did my curiosity! Masters studies gave me the space to dig deeper. I'm here to grow with GenAI-apply improve, and build what's next.
These Days
I'm immersed in the possibilities of language models. Let the big tech handle huge computes while I build end-to-end apps with LLMs, RAG pipelines, and exploring agentic workflows with LangChain, CrewAI and bunch of other tools. I'm also diving deep into post-training techniques like RLHF, PPO, DPO, PEFT and much more. I'm making sure my fundamental concepts are clear.
It's not just about building smarter systems, it's more about making them useful and human-centered.
Achievements I'm Proud of
I've been fortunate to celebrate milestones that shaped me. Being honored with the Chancellor's Gold Medal as the best outgoing undergrad, and winning the Best Research Project Award during my Master's. I've completed impactful internships at Amazon and JP Morgan, authored research papers, contributed to 8+ publications and patents, and built a portfolio of over 10 AI/ML + Full-Stack projects.
Life Beyond Code
I'm always up for a scenic bike ride, and there's something about the ocean breeze and shoreline walks that instantly clears my head. Weekends often mean fresh bakes from the kitchen and the thrill of Formula 1, I love trying out new dessert recipes while cheering on my favorite drivers. Now and then, you'll find me rallying on a badminton court, spinning a foosball table, or kayaking through calm waters.
My work history and internship timeline.
Research Assistant
Currently, with Dr. Shamsad Parvin as the advisor, I have been working on Adaptive Driver Assistance System, that not only detects pedestrians but identfies their intent and behaviour with the help of Low-Rank Adapter modules with Vision Transformer that acts like a plug and play approach.
I led other research projects which are now published like Mapping Crime Dynamics of LA, with an aim to predict crime patterns and areas of concern using Machine Learning algorithms, temporal analysis for number of crimes, and NLP based topic modelling on crime description.
I experimented with 21 different classification and clustering algorithms (XGBoot, Random Forest, DBSCAN, Mini Batch, KMeans, BIRCH etc, name it and you have it!), LSTM for time-series and Latent-Dirichlet Allocation for topic modelling.
Graduate Student Assistant
Collaborated with 200+ students, providing resources and mentorship in Data Intensive Computing projects on machine learning and big data frameworks (Hadoop, MapReduce, Spark) for Fall 2024.
Software Development and Operations Intern
As part of Amazon Pay Core DevOps Engineering Team, I led initiatives to enhance code integrity and operational workflows by rectifying 120+ integration test cases. I worked closely with cross-functional teams which depended extensively on the pipeline, ensuring CI/CD substantially reducing 12 hours of manual efforts per week.
Mitigated 20+ important security risks like breach of confidential data, lack of encryption standards, and automated the process with Python and Java skills for any future instances. This was a major contribution to the team's yearly goal, reducing 10+ different pending actions items.
I tackled the complex task of onboarding a legacy material set for access permissions onto AWS Secrets Manager with S3, IAM, EC2 instances, and Lambda functions. Given the scale of the Amazon Pay application, completely eliminating the dependency on the prior permission handler posed significant challenges, but I successfully managed the transition.
Software Engineering Intern
As part of Consumer & Community Banking (CCB), contributed the team with seamless POC for migration from Splunk to Grafana dashboards, showcasing a 90% effective remodelling strategy. On the frontend, illustrated 5 different data visualizations on Grafana and successfully validated the migration.
I worked backened with an established core JAVA application based on Spring Framework that handles all bank account opening requests for CCB. Redirected requests from Splunk leveraging SoapUI web services, Apache ActiveMQ message requests, Elastic Search configured logs, and Prometheus metric data.
Research Intern
Under the Guidance of Prof. Erik Cambria leading SenticNet Computing Team at NTU, I conducted an extensive research to analyze the errors of sentiment analysis APIs for the company.
With the topic Immigration Reforms Sentiment and Error Analysis and using the SenticNet framework, I analyzed 25K+ tweets for trends and identifying areas for API policy improvements.
Enhanced the performance of 5 different SenticNet APIs, including concept parsing, polarity classification, and subject detection, by comparing against established tools such as Gensim, Vader, and TextBlob.
Here's what sets me apart and makes me unique
Connect with me now!
I'm always looking for new opportunities to connect with people. Share your email and I will ping you back asap.
Don't forget to connect on LinkedIn!
