Skand Tandon

4+ years of work experience

Apr 20251 yr 5 mosPresent

Mar 20232 yrs 1 moMar 2025

Bengaluru, Karnataka, India

• Engineered LoRA based fine-tuning for Hugging Face models using PyTorch, Ray and PEFT • Integrated NVIDIA NIM with K8s solution accelerators, incorporating LLMs, Vector Embeddings, and RAG • Upgraded Apps (Apache Livy and Spark) while maintaining compatibility, improving system reliability. • Developed Helm charts and RBACs increasing efficiency and security for OpenShift • Created Kyverno policies for preemption and optimizing resource for GPU environments. • Enhanced Keycloak based user authentication (OIDC, SAML), enhancing security. • Engineered Spark images with NVIDIA RAPIDS and Whylogs accelerating Spark workloads. • Tuned data pipelines with GPU Direct Storage, boosting stability. • Investigated dynamic PV/PVC solutions, ensuring compatibility, reducing deployment errors. • Enhanced Spark security with audit logs, reducing vulnerabilities by resolving CVEs and boosting compliance. • Consolidated ML pipeline with Superset, Kubeflow, Mlflow, Spark, Airflow, Grafana, EzPresto, Minio S3 • Restructured and maintained a comprehensive suite of Spark examples for customer POCs accelerating client onboarding.

Jun 20222 mosJul 2022

Bengaluru, Karnataka, India

• Analyzed lead acid battery data resulting in a comprehensive dataset that improved battery life predictions by 40%. • Leveraged PySpark to process and analyze datasets, accelerating data processing and increasing data throughput. • Successfully implemented Kalman Filter and LSTM-RNN algorithms to calculate SoH and SoC for batteries.

May 20217 mosNov 2021

Bengaluru, Karnataka, India

Awarded with Certificate of Excellence • Engineered an algorithm for optimal Multi-access Edge Computing server placement, reducing network latency. • Upgraded task scheduling system based increasing resource utilization. • Optimized Quality of Service management in MEC networks, resulting in a decrease in service interruptions.