Headline

Senior Staff Software Engineer | Enterprise GenAI & LLM Inference Optimization on IBM Z (Spyre, AIU, Telum) | Ex-Intel Labs & Grid Dynamics | IIT Madras | IIT Dhanbad

About

I'm currently leading the Z AI Performance team at IBM Research Labs in Bangalore. Our focus is on making generative AI and large language models work reliably at enterprise scale, especially on IBM’s specialized accelerators like Spyre and Telum. One of the projects I’m most proud of is the real-time credit card fraud detection system we built on Telum II for the z17 platform. It now processes 481 billion inferences every day at sub-millisecond latency and was featured in IBM’s commercial launch — directly contributing to revenue growth this year. Seeing something move from an idea in the lab to part of IBM’s flagship hardware was a big moment for the team. Before this, I led GenAI and LLM initiatives at Grid Dynamics for clients including Apple R&D and PepsiCo, delivering multimodal RAG and multi-agent systems with a 100% on-time record. Earlier at Intel Labs, I worked on AI workload optimization using OneAPI tools and helped improve OpenVINO’s NLP capabilities. Those experiences taught me both the technical depth needed to make AI production-ready and the practical challenges of getting it adopted by real teams. What I’ve enjoyed most over the last few years is the leadership side of the work — building and mentoring teams (including groups of research interns), creating internal tools like LogSense that have saved engineers over a thousand hours, and aligning people across locations. I regularly collaborate with teams in the US, Germany, and India. I’ve learned that scaling AI effectively isn’t just about models and accelerators. It’s also about building trust, setting clear priorities, and helping talented people do their best work. I’m at a point where I want to grow as a leader who can connect deep technical work with broader business outcomes and responsible AI practices. I’m particularly interested in opportunities that let me take more ownership of strategy and cross-functional delivery while staying close to the technology. Open to conversations around enterprise GenAI infrastructure, team building, and scaling these systems responsibly.

Location

Bengaluru

AI stack

antigravity

opencode

claudecode

Work experience

Senior Staff Software Engineer at IBM, Aug 2024 - Present · 2 yrs 1 mo, Bengaluru, Karnataka, India, Leading the ISDL Z AI Performance Team, driving strategic Generative AI innovation for IBM Z systems. End-to-end optimization of LLM inference pipelines across WatsonX (Granite), LLaMA, and Mistral on IBM AIU accelerators (Spyre & Telum). Architected and scaled vLLM/llm-d distributed inference for the Spyre accelerator and IBM Triton Inference Server for PyTorch on Telum II, delivering production-grade agentic inference pipelines with industry-leading throughput and sub-millisecond latency (TTFT, TPOT, ITL). Pioneered an enterprise-grade Credit Card Fraud Detection composite AI solution on Telum II (z17), achieving 481 billion fraud detections per day (~5M/sec) at 99.9% sub-millisecond latency using a TabFormer (BERT) + LSTM hybrid architecture. Recognized in IBM z17 GA materials and directly contributing to revenue growth. Built LogSense — a Multi-Agentic AI Log Analysis Platform powered by IBM Bob — reducing manual log triage time by 70% and reclaiming 1,000+ engineering hours annually. Also developed an AI Talent Acquisition Suite that reduced hiring cycle time by 50%. Developed strategic AI roadmaps for IBM Z AI performance initiatives and led cross-functional alignment with global IBM Research teams (US & Germany). Spearheaded technical hiring and org-building to create high-performing teams aligned with enterprise AI growth objectives. Driving AI adoption across IBM ISDL India by identifying high-impact use cases, promoting automation-led productivity gains, and championing responsible AI practices.

Senior Data Scientist at Grid Dynamics, Jan 2024 - Aug 2024 · 8 mos, Hyderabad, Telangana, India, -> Worked as a Senior Data Scientist spearheading in-house LLM-powered GenAI R&D projects with a focus on developing innovative use cases. -> Actively mentored interns and junior team members, accelerating their growth. -> Engaged directly in potential client projects to deliver tailored cutting edge AI solutions.

Software Engineer at Intel Corporation, Dec 2022 - Jan 2024 · 1 yr 2 mos, Hyderabad, Telangana, India, -> As a part of the Intel OneAPI Engineering Team in the Software and Advanced Technology Group (SATG), I develop and implement performance analysis and optimization strategies using Intel VTune, Advisor and other OneAPI tools(Intel Inspector) to improve application performance and reduce development time. -> Led efforts to optimize AI workloads using OneAPI tools like Intel VTune, outperforming Nvidia Nsights by 1.5x performance improvements, reducing model training time for Generative AI models like Bloom, Llama, GANs,Stable Diffusion, other LLMs & other latest models. -> Led & Mentored a team of 4 interns in a project called Deep Learning Profiling, which aims to enhance performance analysis for deep learning applications on VTune using Intel OneDNN, specifically for TensorFlow and PyTorch models. -> Collaborating with other engineers to enable GPU support in OneAPI and optimized performance on Intel GPUs while utilizing Intel Advisor to identify and resolve performance bottlenecks in CPU, GPU, and AI workloads. -> Working closely with customers and partners to understand their needs and provide technical guidance and support in using Intel OneAPI tools to optimize their applications for Intel processors. -> In addition to designing and implementing AI models with Intel OneAPI AI tools, I conducted training sessions for other engineers on how to use Intel Advisor for performance analysis and optimization of their applications. -> As part of the development team, I developed and executed integration, sanity, and unit tests, identified and fixed defects, and implemented test automation scripts using Python and other scripting languages to increase the efficiency and effectiveness of testing. -> Maintained and enhanced test cases and test suites, ensuring comprehensive test coverage across all features of Intel Advisor, and actively participated in code reviews to provide suggestions and feedback on how to improve the software design and code quality.

AI Software Engineer at Intel Corporation, May 2022 - Dec 2022 · 8 mos, Bengaluru, Karnataka, India, -> Spearheaded the development of AI applications, toolkits(Intel OpenVINO), SDKs, and frameworks using cutting-edge techniques in deep learning, computer vision, and edge computing. Optimized performance on various high-performance Intel hardware, including CPUs, GPUs, and VPUs. -> Improved and retrained the natural language processing (NLP) system to enhance feature development and support for low-code and no-code feature enablement. Worked closely with NLP use case teams at Avaamo and TCS to provide continuous support for existing products, launch new IOT products, and integrate NLP activities. Constantly focused on the improvement of use cases with Avaamo SAAS, and translated "English Language + Code (Software)" support in the backend of Avaamo. -> Led the technical development and improvement of various IOTG products, constantly interpreting data to understand and fix issues and failures. Collaborated with cross-functional teams across Intel, Avaamo, and TCS to enhance coding and loading speed resulting in an increase in positive user feedback. Managed multiple client, vendor, and stakeholder relationships, negotiating technical terms, scope, and deadlines.

Software Engineer at Intel Corporation, Jul 2021 - May 2022 · 11 mos, Bengaluru, Karnataka, India, -> As a member of the TD ATA (Technology Development - Advanced Technology Analytics) team, I was responsible for building and deploying scalable AI/ML solutions in various domains including supply chain, semiconductor design, manufacturing, and automation. -> Collaborating closely with the business groups, I analyzed and improved the design of the supply chain, product flows, transactions, and collaborative processes to optimize the end-to-end supply chain performance. -> Working closely with manufacturing engineers, I identified process improvement opportunities and drove the development of 'Factory Scheduling & Decision Support' software solutions to optimize the manufacturing process. -> I was responsible for designing and releasing the TRDI v9.27. This Winform application was deployed in the ATPF (Assembly Test Production Factory) Intel factory to mount reels in an optimized way on a tool that assembles the die onto the substrate. -> In addition to my technical responsibilities, I also played a key role in defining the scope, planning the budget, forming teams, designing and validating solutions, managing program risks, and tracking and communicating program milestones.

Research Intern at Intel Corporation, May 2020 - Jul 2020 · 3 mos, Bengaluru, Karnataka, India, -> As an intern, I contributed to the development of a Constrained Programming (CP) Solver that solved the 'Factory Scheduling Problem' which is currently deployed in Intel Manufacturing factory. My work included using IBM ILOG CPLEX optimization studio, OPL (Optimization Programming Language) & Python Programming language (using doopl interface). -> Additionally, I worked on time optimization for the CP Solver using multiprocessing and multi-threading techniques in collaboration with Intel US Team. Through my efforts, we were able to successfully reduce the overall running time of the solver by around 400 seconds.

Graduate Teaching Assistant at Indian Institute of Technology, Madras, Aug 2019 - Jun 2021 · 1 yr 11 mos, Chennai, Tamil Nadu, India, As a Teaching Assistant for multiple courses, I had the opportunity to collaborate with several distinguished professors and contribute to the academic success of many students. -> Teaching Assistant for CS6910 - Deep Learning, with Prof. Mitesh Khapra -> Teaching Assistant for CS6015 - Linear Algebra & Random Processes, with Prof. Mitesh Khapra -> Teaching Assistant for CS6770 - Knowledge Representation & Reasoning, with Prof. Deepak Khemani -> Teaching Assistant for CS1100 - Introduction to Programming, with Prof. Krishna Moorthy Sivalingam - Responsibilities included creation & grading of students' assignments and exams, helping them practice the material covered in class through weekly office hours, doubt clearing sessions and exam review sessions. * I am proud to have been recognized as the 'Best Teaching Assistant' in the entire CSE department for the academic year 2020-21, and I am grateful for the opportunity to have contributed to the education of future generations.

Education

Master of Technology - MTech, Computer Science, Indian Institute of Technology, Madras

Bachelor of Technology (with Honors) - BTech(Hons), Computer Science, Indian Institute of Technology (Indian School of Mines), Dhanbad