Sayan Chowdhury

Open to work1 years experience

C++ Software Engineer

RemoteTarget Roles: Backend Engineer • Full-Stack Developer • Frontend Specialist

C++ Software Engineer specializing in embedded systems, real-time signal processing, and on-device AI/ML deployment.

Standing Rank

Rank Not Available

Developer Badges

No badges earned yet

Skills & Technologies

50 skills

c++cpythonmultithreadingmemory leak analysisruntime stabilityembedded softwaretcncnnbertdeep learningmodel training & deploymentspectrogram analysislarge language models (llms)llm fine-tuningprompt engineeringrag systemsagentic ailangchainlanggraphlangfuseworkflow automationtensorflow litencnnnntraineron-device inferencemodel conversion (onnx)quantization (int8/fp16)knowledge distillationself-distillationpytorchtensorflowkerastizen os (linux)androidqualcomm hdk-8750rtos-adjacentgitperforcegdbadbvisual studio codelinuxbuild serversdsaoperating systemsoopcomputer networkstcp/ipmysql

GitHub Activity Evidence

GitHub Footprint Not Connected

Recruiters value verified code contributions. Linking a GitHub account displays live heatmaps, active contribution metrics, repository languages, and push activity tags.

Work Experience

Software Development Engineer

Samsung Research Institute

Jul 2025 - Present

Developed multithreaded C++ modules for a real-time ultrasound proximity detection system on Qualcomm HDK-8750, implementing IQ demodulation, CIC filtering, and phase unwrapping; improved re-detection accuracy by 35%.
Engineered a real-time audio classification system for snoring and 8-class environmental sound event detection using a TCN student distilled from MobileNetV1; boosted F1-score by 12%, cut model size 60% (14 MB to 5.6 MB), and reduced peak memory by 35% on Samsung AI Lite chips with under 1% accuracy degradation.
Served optimized audio ML models as low-latency REST inference endpoints using FastAPI, achieving average response times under 50 ms and integrating outputs into downstream agentic pipelines for autonomous smart-home alerting.
Evaluated 3+ multimodal audio-language models (Audio Flamingo) across 10+ real-world acoustic scenarios; applied generative AI zero-shot prompting strategies, cutting manual annotation effort by an estimated 40%.
Optimized deep learning models for edge deployment via TensorFlow Lite, NCNN, and NNTrainer; achieved 3x faster inference through INT8/FP16 quantization, operator fusion, and graph-level optimizations across ONNX and PNNX formats.
Prototyped a device-free human activity recognition system using LSTM models on WiFi Channel State Information (CSI) signals to classify human activity patterns without wearable sensors.
Reduced production crash rate and resolved complex multithreaded C++ stability issues using GDB, memory profiling, and log analysis; built unit tests and automation scripts ahead of commercialization.

Projects

AI-Powered Rental Pricing Tool (Real Estate)

PythonLLM AgentEmbedding-based Similarity SearchGeospatial Data Integration

Built an AI-assisted rental pricing engine combining property attributes, historical rental data, and external location signals (school quality, road noise, flood risk, amenities) to recommend comparables and a fair price.
Architected an NL feedback loop using an LLM-based conversational agent and embedding-based similarity search, letting users adjust comparables and persist learnings for future pricing accuracy.

Enhanced BERT Transformer for Faster Text Classification

PythonPyTorchJupyter NotebookGoogle Colab

Optimized BERT's attention mechanism in PyTorch, improving inference speed for real-time text classification tasks.

Memory Management Technique (MMT) Simulation

C++Visual Studio Code

Built a C++ demand paging simulator with dynamic page loading and FIFO/LRU replacement algorithms for efficient memory management.

Leaderboard Standings

Leaderboard Position Pending

Global test scores, peer standing percentiles, and algorithm leaderboard ranks are updated dynamically.

Assessment Highlights

Assessments Not Completed

Coding evaluations, system assessment results, and conceptual score badges will appear here after taking a test.

AI Collaboration Score

AI Collaboration Score Pending

Developer coding behavior, assistant cooperation, and AI pair-programming indicators are evaluated during live coding sessions.

Role Compatibility Profile

Role Compatibility Analysis Pending

Custom matching reports, candidate role compatibility percentiles, and core engineer strength profiles are processed once conceptual code screenings are complete.

Achievements

Best New Tech Award

Best New Tech Award - Samsung (2026)

Samsung Professional Exam

Samsung Professional Exam - Cleared by top 10% of all Samsung employees globally (2025)

About Details

Professional Bio

C++ Software Engineer specializing in embedded systems, real-time signal processing, and on-device AI/ML deployment. Experienced in building production-grade multithreaded C++ software and optimizing deep learning models for edge devices.

M.Tech in Computer Science and Engineering in Computer Science and Engineering

National Institute of Technology Warangal (2023 - 2025)

B.Tech in Computer Science and Engineering in Computer Science and Engineering

Jalpaiguri Government Engineering College West Bengal (2019 - 2023)

Languages: English, Hindi