Open to opportunities

Raj Srujan Jalem

@rajsrujanjalem

Message

I build and scale low-latency ML platforms across edge and cloud environments.

India

Message

What I'm looking for

I'm looking to lead teams building reliable, high-performance ML platforms, distributed inference systems, and production AI infrastructure across edge and cloud environments.

At Vimaan, I architect the core ML infrastructure and lead a team of nine engineers building high-performance Edge AI systems. My pipeline optimization reduced inference latency by 70% while processing 1–2 TB of multi-camera video data each day.

I've built distributed inference systems with ROS, Ray, Triton Inference Server, TensorRT, and dynamic batching, addressing edge bandwidth, packet loss, GPU cost, and ultra-low-latency requirements. I also created an LLMOps observability stack with OpenSearch for distributed logging, microservice health monitoring, and automated root-cause analysis.

Previously, I reduced annotation costs by about 40% at Alectio, cut API latency by 70% and accelerated deployments 3x at Fasal, and improved computer-vision throughput by 50% at Utopia Global. I enjoy taking AI platforms from prototype to reliable production systems while mentoring engineers and leading cross-functional teams.

Experience

Work history, roles, and key accomplishments

Current

Manager, MLOps

Current

Vimaan

Aug 2022 - Present (4 years 1 month)

Architected core ML infrastructure and managed a team of 9 engineers, achieving a 70% reduction in inference latency. Built a distributed, on-premise Edge AI pipeline using ROS streaming to process 1-2 TB of daily multi-camera video data.

MLOps Edge AI ROS Ray Triton Inference Server TensorRT OpenSearch Kubernetes Docker Python C++Wireshark

ML Backend Engineer

Alectio

Jan 2022 - Aug 2022 (7 months)

Developed the Alectio Model Library to standardize model integration, automated training, and active-learning workflows. Built automated labeling pipelines that reduced customer data annotation costs by ~40%.

Backend Development Microservices Active Learning Automated Labeling Python

AI Data Engineer

Fasal

Mar 2021 - Jan 2022 (10 months)

Set up automated forecasting and continuous training workflows on Kubernetes using Kubeflow and MLflow. Migrated monolithic pipelines to a microservice architecture and implemented a Feature Store, reducing API latency by 70%.

Kubernetes Kubeflow MLFlow Microservices Feature Stores CI CD Docker

Data Science Engineer

Utopia Global

Jan 2019 - Mar 2021 (2 years 2 months)

Trained custom U-Net models for super-resolution image enhancement, doubling the output resolution quality. Refactored production vision pipelines to increase processing throughput by 50%.

U Net SuperResolution Computer Vision NLP NER EC2 Python

Staff MLOps Engineer

Vimaan

Resolved edge bandwidth bottlenecks via a dynamic frame-sampling module, using upstream object detection to trigger selective 4K image acquisition. Tuned ROS 1 and ROS 2 network configurations using Wireshark to diagnose packet loss, ensuring stable, high-throughput video streaming across edge nodes.

ROS ROS2 Wireshark Edge AI Object Detection Image Data Acquisition Network Configuration

Senior MLOps Engineer

Vimaan

Scaled server-side inference using Ray and Triton Inference Server, applying TensorRT quantization and mixed-precision optimization alongside dynamic batching to cut GPU compute cost while sustaining ultra-low latency. Built an LLMOps observability stack using OpenSearch to aggregate distributed logs, monitor microservice health, and automate root-cause analysis.

Ray Triton Inference Server TensorRT Mixed Precision Dynamic Request Batching OpenSearch LLMOps Observability