> About Me
AI Profiling Engineer at IBM, benchmarking LLMs and profiling the IBM Spyre AI accelerator. Background spans 3+ years across big data engineering, GPU computing, and deep learning.
I dig into device-side and CPU operations to surface architectural overheads and performance bottlenecks, and love building scalable data pipelines and ML systems. Holds an MSc in AI & ML from the University of Limerick.
> Core Attributes
> Skill Tree
Unlocked abilities and proficiencies
Languages
MasterAI / Machine Learning
ExpertBig Data
ExpertCloud & Tools
Advanced> Interactive Skill Tree
Click to unlock new abilities
Python Mastery
Advanced Python programming
PyTorch Expert
Deep learning with PyTorch
CUDA & GPU Computing
GPU-accelerated kernels
AI Profiling
IBM Spyre & PyTorch profiling
LLM Benchmarking
PT-Util & bottleneck analysis
Accelerator Architecture
Next-gen AI silicon
> Quest Log
Career journey and achievements
AI Profiling Engineer
International Business Machines (IBM) • Waterford, Ireland
Nov 2025 – Present
- ▸Benchmarked LLMs such as Qwen and Granite for the IBM Spyre performance benchmarking suite, analyzing device-side and CPU operations and calculating PT-Util metrics to identify architectural overheads and bottlenecks
- ▸Developed profiling test cases in the torch-spyre repository to measure CPU time, kernel execution time, and memory usage; integrated upstream PyTorch profiling tests for Torch 2.11
- ▸Conducted comparative analysis of GPU architecture performance across 17,000+ profiling metrics spanning memory bandwidth, compute utilization, and interconnect performance
Big Data Analyst
Cognizant Technology Solutions • Chennai, India
Jul 2022 – Aug 2024
- ▸Designed and maintained real-time data pipelines using Kafka, Apache Spark, and PostgreSQL to reduce operational latency and enable low-latency analytics
- ▸Engineered data ingestion workflows that accelerated data availability, leading to a 15% improvement in reporting time
- ▸Integrated Grafana dashboards with Spark outputs to visualize streaming KPIs, improving decision-making cycles across business units
Cloud Computing Intern
Cognizant Technology Solutions • Chennai, India
Jan 2022 – Jun 2022
- ▸Developed cloud pipelines routing API data into Azure Data Lake Storage, enhancing data availability across the analytics platform
- ▸Engineered views and external tables in Synapse Analytics to accelerate query performance
MSc in Artificial Intelligence and Machine Learning
University of Limerick • Ireland
Sept 2024 – Aug 2025
- ▸QCA: 3.58/4.00
- ▸Modules: AI, Deep Learning, Data Pipelines, Reinforcement Learning
BE in Electronics and Communication Engineering
St. Joseph's College of Engineering • Chennai, India
Aug 2018 – Jun 2022
- ▸GPA: 8.10/10
> Completed Quests
Major projects and achievements
GPU-Optimized Computation using CUDA
Developed and benchmarked CUDA programs to accelerate fundamental operations: 1D convolution using constant memory, element-wise vector addition with CUDA threads, and matrix multiplication using 2D thread blocks to leverage GPU cores.
CLI-Based Game Development using Rust
Designed a command-line game in Rust following SOLID principles and OOP patterns: Factory Method for player creation, State pattern for game states, Observer for event tracking, and Memento for save/restore serialization.
Deep Reinforcement Learning for Atari (Air Raid)
Engineered a Deep Q-Network (DQN) agent in TensorFlow to master the Atari game Air Raid. Preprocessed frames with Gym wrappers into a CNN policy network, using experience replay, target network updates, and an epsilon-greedy strategy.
Evolutionary Algorithm Feature Selector
Built a genetic algorithm-based wrapper feature selector using DEAP and scikit-learn. Tuned evolutionary hyperparameters to reduce feature set size by 50% while preserving accuracy and improving F1-score.
> Achievements Unlocked
Certifications and milestones
Metrics Marathon
Analyzed 17,000+ GPU profiling metrics across bandwidth, compute, and interconnect
Upstream Contributor
Integrated PyTorch 2.11 profiling tests into the torch-spyre repository
LLM Benchmarker
Benchmarked Qwen and Granite LLMs on the IBM Spyre accelerator
MSc Distinction Track
MSc in AI & ML at University of Limerick — QCA 3.58/4.00
Reporting Accelerator
Improved reporting time by 15% with optimized data ingestion workflows
Efficiency Evolver
Cut feature set size by 50% via genetic algorithms while preserving accuracy
> Active Quests
Complete challenges to earn rewards
Explore the Portfolio
Scroll through all sections
Master the Skills
View all skill categories
Connect with Me
Click on a social link
Achievement Hunter
Discover all achievements
> Start a Quest Together
Ready to collaborate? Let's connect!
Contact Information
Current Status
Open to roles in AI Performance Engineering, ML Engineering, and Data Engineering