Aryan Kasat

Data Scientist

Pune, Maharashtra, India2 yrs 2 mos experience

Most Likely To SwitchAI ML Practitioner

Key Highlights

Expert in AI applications including NLP and Generative AI.
Proven track record in developing scalable AI solutions.
Strong foundation in Machine Learning and Data Engineering.

Stackforce AI infers this person is a Data Scientist specializing in AI solutions across various industries.

Contact

Skills

Core Skills

Artificial Intelligence (ai)Machine Learning

Other Skills

Analytical SkillsAutogenBack-End Web DevelopmentC (Programming Language)C++Cascading Style Sheets (CSS)Collaborative Problem SolvingCommunicationCreative Content CreationDeep Q-learningEntrepreneurshipExploratory Data AnalysisFP-Growth AlgorithmFastAPIFinancial Accounting

About

I'm a Data Scientist with a deep passion for research-driven AI and how it can actually be applied to solve real-world business problems across different domains. I’ve developed a strong appreciation for the role research plays in shaping the AI landscape—especially when it comes to building scalable, impactful solutions. I’ve worked on a variety of AI applications including NLP, Generative AI, LLMs, and even some multi-modal applications like Diffusion Language Models. Long term, I see myself growing into a Solutions Architect role. I’m also embarking on the journey to expanding my existing knowledge in Data Engineering and MLOps so I can better contribute to building end-to-end AI systems—from prototype to production.

Experience

2 yrs 2 mos

Total Experience

1 yr 1 mo

Average Tenure

1 yr 7 mos

Current Experience

Tata consultancy services

Data Scientist @ AI CoE

Nov 2024 – Present · 1 yr 7 mos · Pune, Maharashtra, India · On-site

1. Extracted data from reports using OCR for unsupervised fine-tuning of SLMs and generated synthetic data using LLM across five categories for supervised fine-tuning of SLMs for making the SLMs domain-adaptive.
2. Fine-tuned SLMs using Masked Language Modeling (MLM) and LoRA techniques, followed by evaluation using LLM-as-a-judge methodology to evaluate model performance.
3. Upgraded existing RAG pipelines to Agentic-RAG by integrating multiple external data sources using Model Context Protocol (MCP) through Autogen framework enhancing end-user outcomes.
4. Designed and developed Agentic AI Cards using the Semantic Kernel framework, through multi-agent collaboration to address complex business use cases.

OCR - TessaractLLMsModel Context ProtocolAutogenMasked Language ModelingLoRA+4

Visulon inc.

Data Science and AI/ML intern

Jun 2023 – Aug 2023 · 2 mos · Pune, Maharashtra, India · On-site

Visulon has a product Assortment Planner. Improvising this product, built a feature to recommend products to assortment planners in real time. Used the following algorithms for the complete implementation of the model.
1. Developed a Popularity Index Algorithm for cold-start users.
2. Used K-Modes Clustering Algorithm for user previously conducted assortment planning.
3. Used FP-Growth Algorithm for similar product recommendations to the product chosen by the user during assortment planning.
Performed MVP as well as an end-to-end developed product along with live demonstration
of the model to the clients.

Popularity Index AlgorithmK-Modes Clustering AlgorithmFP-Growth AlgorithmMachine LearningArtificial Intelligence (AI)

Ihub divyasampark @ iit roorkee

Research Fellowship

Oct 2022 – May 2023 · 7 mos

Building an intelligent traffic management and engagement system reduces a vehicle's waiting time at a traffic junction by nearly 35%-40%.It would help decrease pollution significantly.
1. Used Deep Q-learning as our Neural Network and Deep Reinforcement model to train on data.
2. The state space used to detect the traffic at a signal junction is the q-length of vehicles in each lane.
3. The reward function is the waiting time of each vehicle in the lane.
Used Simulation of Urban Mobility (SUMO) environment to train the model and check its working.

Deep Q-learningSimulation of Urban Mobility (SUMO)Artificial Intelligence (AI)Machine Learning

Microsoft

Engage Mentee

May 2022 – May 2022 · 0 mo · India

Developed Data Analysis application using automobile industry dataset.I have used
Machine Learning models for doing:
1. Exploratory Data Analysis
2. Linear Regression for price prediction of car
3. K-means clustering for Customer Segmentation.

Exploratory Data AnalysisLinear RegressionK-means clusteringMachine Learning