Headline

CS @ PES University | Computer Vision & MLOps | PyTorch · TensorRT · Edge AI (Jetson) | IEEE Co-Author | Seeking New-Grad ML/CV Roles

About

Graduate of PES University, focused on Computer Vision and MLOps — I build deep learning systems end to end, from raw data engineering through to optimized inference on edge hardware. As an AI Intern at Inferigence Quotient, I worked across the full vision pipeline: large-scale dataset engineering (COCO→DOTA conversion, Oriented Bounding Box generation), production image-tiling engines for OBB detection, automated annotation refinement with Segment Anything (SAM), and model interpretability via GradCAM/GradCAM++. On the deployment side, I built an edge-benchmarking toolkit for NVIDIA Jetson — parsing Tegrastats and TensorRT logs to profile P50–P99 latency, power, and utilization across FP16/FP32/INT8 precision modes. Earlier, as Lead Software Developer on an autonomous pest-detection rover at the PESU Center for IoT, I deployed a CNN-based vision system on Raspberry Pi and ESP32 with automated drift monitoring. That work was co-authored into a research paper presented at an IEEE conference. Core toolkit: PyTorch · TensorFlow · OpenCV · YOLO/OBB · TensorRT · NumPy · Shapely/NetworkX · FastAPI · Linux/Git I'm looking for new-grad roles in computer vision, ML engineering, or MLOps where I can own problems end-to-end. Open to connecting — akilanelango2004@gmail.com.

Location

Bengaluru

AI stack

claude code

claude

gemini

openai

claudecode

ChatGPT

Work experience

AI Intern at Inferigence Quotient Pvt Ltd, Jan 2026 - May 2026 · 5 mos, Bengaluru, Karnataka, India, - Built an end-to-end COCO→DOTA dataset pipeline: filtered large-scale datasets by class and converted polygon masks to minimum-area Oriented Bounding Boxes (OpenCV + NumPy) at scale. - Developed production image-tiling engines for OBB detection — configurable overlap tiling, Shapely-based polygon intersection/area filtering, and full directory mirroring. - Engineered an automated annotation-refinement pipeline using Segment Anything (SAM) to tighten labels from model masks, with YOLO-OBB output and per-image metadata traceability. - Integrated GradCAM/GradCAM++ into mmYOLO workflows and built a quantitative heatmap-analysis framework (per-class TP/FP/FN correspondence, coverage ratios) producing diagnostic CSVs. - Built an NVIDIA Jetson edge-benchmarking toolkit: custom Tegrastats/TensorRT log parsers and a telemetry suite computing P50–P99 latency, 1ms-binned histograms, and power/utilization across FP16/FP32/INT8. - Collaborated on the full ML lifecycle (dataset prep → training → evaluation → edge deployment) in Linux/Git environments.

Intern at PESU Center For IoT, Jun 2025 - Jul 2025 · 2 mos, Bengaluru, Karnataka, India, - Led software development of an autonomous agricultural rover for real-time pest detection. - Designed and deployed a CNN-based vision system on Raspberry Pi and ESP32 for on-device edge inference. - Built automated monitoring for schema and statistical data-drift detection. - Co-authored a research paper presented at an IEEE conference (Jan 2026).

Education

Bachelor of Technology - BTech, Computer Science, PES University

High School Diploma, CBSE, The Brigade Schools - Bengaluru

Public links

github: https://github.com/akilanelango