Lê Chí ĐạiAbout
I’m Lê Chí Đại, an AI engineer at Rainscales and a Computer Science Honors undergraduate at HCMUT. I deploy vision models on edge GPUs, build evaluated LLM applications, and want to specialize in LLM inference serving.
How I work
I take ownership of core implementation and coordinate how the pieces fit together. In team projects, my work spans architecture, AI and backend integration, and bringing components into a working prototype. My research adds experience with model experiments and evaluation.
CoinSight shows how I build and evaluate LLM agents; VoiceReaderAIshows how I connect AI services to an application and lead integration.
Experience
AI Engineer, Rainscales · Jun 2026–present
Real-time AI detection. I deploy detectors as FP16 TensorRT engines on NVIDIA Triton with dynamic batching and run DeepStream on live RTSP feeds across 15 rig fleets, up to 30 concurrent streams per Jetson Orin. I built a second-stage verifier that sends cropped multi-frame evidence to a self-hosted Qwen3-VL, raising alert precision from 6.6% to 18.7% on the test set, and extended red-zone rules to 360° fisheye cameras with ego-motion compensation.
AI Intern, AKTech · Apr–Jun 2026
Marker-free multi-camera calibration. I co-developed a service that calibrates multi-camera rigs from a person’s T-pose instead of a checkerboard, implementing pose-based RANSAC initialization and confidence-weighted Ceres bundle adjustment.
Research
Since 2025, I’ve worked with ML4U Lab & AISIN on monocular 3D object detection. My work includes conditioning detectors on camera geometry through camera-intrinsic embeddings, and investigating Mixture-of-Experts for generalization across sensor domains.
The camera-intrinsic embedding work improved AP_3D Hard on KITTI from 20.80 to 21.13. I’m also researching joint multi-dataset detection across KITTI, nuScenes, and Waymo.
Q. H. Nguyen, C. D. Le, et al. “MonoBEN: Monocular 3D Object Detection Beyond Fixed Camera iNtrinsics.” RIVF 2026 (IEEE), accepted for presentation.
Education
B.S. Computer Science, Honors Program
Ho Chi Minh City University of Technology
Expected graduation: 2027 · GPA: 3.9/4.0
Merit scholarships for five consecutive semesters, SEM231–SEM251.
TOEIC 905.
Other work
At the NAVER Vietnam AI Hackathon in 2025, I led the technical architecture, core AI/backend implementation, and integration ofVoiceReaderAI. Our four-person team built an Android reading assistant combining CLOVA OCR and Voice with Gemini, LangChain, and Pinecone for narration, summaries, and contextual Q&A, with Vietnamese voice cloning through VieNeu-TTS. We were selected as national finalists.
For ZALO AI Challenge 2025, I worked on synthetic data generation for few-shot aerial object detection, using SAM to extract objects and compose augmented training scenes.