Sribharath Kainkaryam

<p>Machine Learning and Data Engineering for commodity trading<br>
Developed a distributed and fault-tolerant scraper framework for gathering data from various sources, optimized resource utilization<br>
Designed and deployed processes to ensure data reliability by setting and checking data quality standards, ensuring significant error reduction in data processes<br>
Leading the design for cloud based integrated data analytics platform for firmwide usage, catering to data engineers and analysts<br>
Technologies used: Apache Airflow, Kubernetes, Docker, Terraform</p>

• Developed a distributed and fault-tolerant scraper framework for gathering data from various sources, optimized
resource utilization
• Designed and deployed processes to ensure data reliability by setting and checking data quality standards, ensuring
signficant error reduction in our data processes .
• Leading the design for cloud based integrated data analytics platform for firmwide usage, catering to data engineers
and analysts
• Technologies used: Apache Airflow, Kubernetes, Docker, Terraform

<p>Machine Learning and High Performance Computing for energy prospecting<br>
Incubated, led development and open-sourcing of mdio, a cloud native, scalable storage engine for various types of energy data<br>
Drove growth and commercialization of a SaaS solution for data management<br>
Led to savings of ∼$x0 MM per year in seismic data storage cost for hundreds of petabytes<br>
Grew the data science team from first hire to scaling it by 8x in four years, driving strategic initiatives, and subsequently promoted to team lead role<br>
Collaborated with HPC to design and implement a cost-aware hybrid (on-prem + cloud) high-performance computing roadmap, while establishing robust software engineering practices<br>
Provided technical evaluation of vendor proposals and worked with vendors to build components of data engineering infrastructure<br>
Co-authored a book on application of machine learning in oil and gas<br>
Designed and developed scalable microservices for data streaming and inference<br>
Key technologies used: FastAPI, gRPC, Protocol Buffers, Flask, AWS API Gateway<br>
Developed elastic, fault tolerant, distributed computing infrastructure for scaling up training and inference of deep neural networks on terabytes of seismic data<br>
Key technologies used: PyTorch, torchelastic, Kubernetes, Docker, AWS SageMaker, Vertex AI, MLFlow<br>
Lead for research and development of SaltNet, a workflow for interpreting salt body for velocity model building in seismic imaging<br>
Key technologies used: PyTorch, Dask, TensorRT, Docker<br>
Formulated the problem and designed the dataset for TGS Kaggle salt identification challenge, subsurface exploration industry’s first open data challenge on Kaggle<br>
Co-authored several conference abstracts and journal papers, awarded best paper in The Leading Edge, a premier exploration geophysics journal for 2020</p>

<p>Research and high-performance software development for massively parallel seismic applications<br>
Identified and proposed solutions to workflow inefficiencies in velocity model building workflows costing $1.5 million per year<br>
Lead developer for TraceRay – Schlumberger’s proprietary object-oriented ray tracing library (∼150,000 lines of code) – designed APIs, consulted with users of the library and provided technical support<br>
Achieved a 20% improvement in computational efficiency of TraceRay by redesigning core algorithms and numerical methods for ray tracing</p>

<p>SaaS product for radiation oncology<br>
Mentor on using Apache Airflow for automating data pipelines, improving efficiency in tasks like data preprocessing and model training<br>
Guide integration of MONAI for tasks such as image segmentation, enhancing accuracy in organ and tumour contouring and treatment planning<br>
Advise on data annotation techniques and scalable model deployment, improving accessibility of AI solutions in radiation oncology<br>
Technologies used: MONAI, MONAILabel, NVIDIA Triton</p>