ML Engineer · IIoT Specialist · Embedded Linux

Nilay N.
Jaltare

// Edge AI · Industrial IoT · AWS Cloud · Linux Systems
// NVIDIA Jetson · Raspberry Pi · MQTT · LSTM · Transformer

I build intelligent systems that live at the edge — deploying optimised ML models on embedded Linux hardware, engineering low-latency MQTT pipelines, and backing everything with production-grade AWS cloud infrastructure. Currently leading industrial leak detection for live crude oil pipelines at Bharat Flow Analytics.

NVIDIA Jetson Nano Raspberry Pi Jetpack SDK AWS IoT Core MQTT/TLS EDA Linux Admin CPU Affinity Shell Scripting
45%
Latency Reduction
99%
Transmission Reliability
80%
Faster OTA Deployment
35%
Processing Speed Gain
🏅
AWS Certified Machine Learning Engineer Associate · 2025 · Validated Production ML on AWS
00

Target Focus Areas

01 / 03 🧠
Artificial Intelligence & Machine Learning
Edge AI · Time-Series · Anomaly Detection

Hands-on experience building and deploying LSTM, Transformer, and Autoencoder architectures for real-time anomaly detection on live industrial sensor data. Trained models are optimised for low-power edge hardware using TensorFlow and PyTorch, with production deployments on NVIDIA Jetson platforms running embedded Linux. AWS SageMaker and Bedrock experience rounds out the cloud-side ML stack.

  • ML Engineer — Edge AI / Inference Optimisation
  • Applied Scientist — Industrial / Predictive Maintenance
  • AI Platform Engineer — AWS / SageMaker
  • Computer Vision Engineer (OpenCV, PyTorch)
02 / 03 🏭
Industrial IoT & Cloud Architecture
IIoT · MQTT · AWS IoT Core · EDA

Built production IIoT systems for crude oil pipeline monitoring — end-to-end: from UART sensor interfaces and MQTT/TLS pipelines through AWS IoT Core, Lambda, DynamoDB, and S3, to real-time Gunicorn-served dashboards. Implemented Event-Driven Architectures achieving sub-10ms event propagation and OTA firmware updates eliminating the need for manual field visits.

  • IIoT Platform Engineer
  • Embedded IoT Systems Engineer
  • Cloud IoT Architect — AWS
  • OT/IT Integration Engineer
03 / 03 ⚙️
Embedded Linux & Systems Engineering
Linux · Jetson · RPi · Shell · Systems

Deep practical Linux experience on NVIDIA Jetson Nano (Jetpack SDK, L4T) and Raspberry Pi — CPU affinity tuning, dynamic frequency scaling, real-time process management, serial communication (UART/AT), and automated system monitoring via shell scripting. Comfortable at the hardware-software boundary: from PCB electrical spec alignment to embedded firmware and kernel-level system optimisation.

  • Embedded Linux Engineer — NVIDIA, Qualcomm, NXP
  • Systems Software Engineer — Edge Platforms
  • Firmware / BSP Engineer — Jetson / ARM
  • Linux Platform Engineer — Robotics / Autonomous Systems
01

About

I'm a Machine Learning and IoT Engineer (SDE-I) at Bharat Flow Analytics, Pune, where I lead the design, deployment, and client delivery of ML-powered leak detection systems for live crude oil and utility pipeline operations across India.

My work lives at three intersections: embedded Linux systems (NVIDIA Jetson Nano, Raspberry Pi), ML model engineering (LSTM, Transformer, Autoencoder architectures), and cloud-backed IIoT infrastructure (AWS IoT Core, Lambda, DynamoDB, SNS). I've tuned systems at the kernel level — CPU affinity for core-specific task allocation, dynamic frequency scaling, and Linux-based automation scripts for real-time metrics monitoring.

On the Linux side, I'm proficient in system administration, shell scripting, log analysis with grep/regex pipelines, process and performance management, and embedded system configuration on Jetpack SDK. My systems run on field hardware in conditions where reliability is non-negotiable — a context that forces precision.

I mentor engineering interns, coordinate with third-party PCB hardware vendors, and manage the full product lifecycle from technical consultation through field trial. I hold a B.Tech in ECE with an AI/ML minor from MIT World Peace University, and am AWS Certified in Machine Learning.

Location Pune, India
Current SDE-I · Bharat Flow Analytics
OS Linux (primary), Ubuntu, L4T/Jetpack
Hardware NVIDIA Jetson Nano · Raspberry Pi
Cloud AWS · IoT Core · SageMaker · Bedrock
B.Tech ECE + AI/ML Minor · MIT WPU · 2025
Lang English · Hindi · German (B1)
02

Linux & Systems Proficiency

nilay@jetson-nano:~$
# CPU affinity — pin ML inference to cores 2-3 $ taskset -cp 2-3 $(pgrep inference_engine) pid 3847's current affinity list: 0-3 pid 3847's new affinity list: 2-3   # Dynamic frequency scaling — max performance governor $ sudo jetson_clocks --show CPU: online=0-3 governor=performance freq=1479000 GPU: 921600000 Hz EMC: 1600000000 Hz   # Real-time system metrics monitor script $ ./monitor.sh --interval 500ms --log /var/log/telemetry [00:00:01] CPU=38% MEM=62% TEMP=43°C MQTT=UP [00:00:02] CPU=41% MEM=63% TEMP=44°C MQTT=UP   # Log analysis pipeline with grep + regex $ grep -E "ANOMALY|ALERT" /var/log/pipeline.log \ | awk '{print $1, $2, $NF}' \ | sort | uniq -c | tail -20 3 2025-04-12 08:14:02 ANOMALY:pressure_spike 1 2025-04-12 08:14:09 ALERT:threshold_breach   # UART serial interface — AT command validation $ stty -F /dev/ttyUSB0 9600 raw && echo "AT" > /dev/ttyUSB0 OK ← device handshake confirmed
Process & Performance Management PRODUCTION EXPERIENCE
CPU Affinity (taskset) Dynamic Freq Scaling jetson_clocks htop / top nice / renice systemd services cgroups Process scheduling
Shell Scripting & Automation PRODUCTION EXPERIENCE
Bash scripting grep / regex awk / sed cron / crontab Log analysis System monitoring scripts Pipe & redirection xargs
Embedded Linux / Jetpack / BSP PRODUCTION EXPERIENCE
Jetpack SDK L4T (Linux for Tegra) NVIDIA Jetson Nano Raspberry Pi OS GPIO / I2C / SPI UART / AT Commands Serial comms (tty) OTA updates
Networking & System Admin PROFICIENT
MQTT/TLS systemd-networkd iptables / UFW SSH / rsync Filesystem management Package management (apt) Docker on Linux User/permission mgmt
03

Hardware Platforms & Embedded Experience

🟢
NVIDIA Jetson Nano
ARM Cortex-A57 · 128-core Maxwell GPU · Jetpack SDK
  • Deployed real-time ML inference for leak detection anomaly classification on live pipeline sensor data.
  • CPU affinity-based task allocation across 4 cores — 25% workload efficiency gain.
  • Dynamic frequency scaling via jetson_clocks35% processing speed improvement.
  • Developed and maintained MQTT/TLS pipelines to AWS IoT Core with sub-10ms event propagation.
  • Linux-based scripting for real-time acquisition, process monitoring, and automated alerting.
  • Configured Jetpack SDK, L4T runtime, and CUDA-accelerated OpenCV for edge vision pipelines.
🔴
Raspberry Pi
ARM Cortex-A72 · Raspberry Pi OS (Debian Linux)
  • Designed and deployed a robust edge data logger monitoring critical industrial parameters in the field.
  • Implemented secure MQTT/TLS pipeline to AWS IoT Core achieving 99% transmission reliability.
  • Architected OTA firmware update system — 80% reduction in firmware deployment time, eliminating manual field visits.
  • Developed Python automation scripts for edge-side data acquisition, local anomaly detection, and DynamoDB/S3 logging.
  • Integrated AWS SNS alerting, improving incident response time by 40%.
  • Configured GPIO/UART interfaces for hardware sensor integration and AT command validation.
🔌
PCB & Hardware Integration
Embedded Hardware-Software Interface
  • Technical point-of-contact for third-party PCB manufacturers — aligning electrical design specs with embedded software requirements.
  • Serial communication (UART / AT commands) for sensor data acquisition from field devices.
  • Payload parser development and device shadow management in AWS IoT Core for sensor state synchronisation.
  • Field hardware deployment on live crude oil pipelines — resolved false-alarm anomalies through algorithmic and hardware-level refinements.
  • Experience with signal conditioning, sensor calibration, and real-time data acquisition.
04

Experience

Software Development Engineer – I
Bharat Flow Analytics · bharatflow.in
Jun 2025 – Present Pune, India

ML model design, cloud-integrated dashboards, and client-facing technical delivery for industrial leak detection across the organisation's product portfolio.

Machine Learning & Systems
  • Time-Series Models: Developed and validated LSTM and Transformer-based classification models on telemetry datasets — significant improvement in precision and F1 score for leak detection accuracy.
  • Cloud Architecture: End-to-end monitoring stack on AWS (EC2, S3, SNS) ingesting edge sensor data with Gunicorn-served real-time analytics dashboard. ~45% cost savings
  • Desktop Analytics: Cross-platform Tkinter/PyQt visualisation apps packaged as standalone executables for offline field engineering. +20% UX score
  • Industrial Deployment: Led transition from test to live crude oil pipeline operations — resolved false alarm anomalies through algorithmic refinement.
Embedded Linux & Leadership
  • Project Leadership: Managed full product lifecycle from concept to field trial — coordinating with client stakeholders, hardware vendors, and intern teams simultaneously.
  • Linux Systems: Continued ownership of embedded Linux configurations on Jetson Nano deployments — process management, system monitoring scripts, and MQTT pipeline stability under field conditions.
  • Hardware Liaison: Technical point-of-contact for third-party PCB firms — aligning electrical design with embedded firmware requirements.
  • Intern Mentorship: Guided engineering interns on ML model development, IoT integration patterns, and Linux-based product engineering best practices.
Engineering Intern
Bharat Flow Analytics
Jun 2024 – Jun 2025 Pune, India

Deployed scalable, cloud-connected IoT solutions contributing to a production-ready leak detection system for oil and utility pipelines.

Edge & IoT Engineering
  • Validated and deployed ML models for leak detection on edge devices. +30% system reliability
  • Designed Event-Driven Architecture (EDA) using AWS IoT triggers and Lambda. -45% latency
  • MQTT pipelines between NVIDIA Jetson Nano and AWS IoT Core with sub-10ms event propagation.
  • Built custom payload parsers and device shadows in AWS IoT Core with Lambda, DynamoDB, and S3 for real-time processing and long-term storage.
Linux Systems Optimisation
  • Implemented CPU affinity (taskset) for core-specific ML inference task allocation on Jetson Nano. +25% workload efficiency
  • Applied dynamic frequency scaling via jetson_clocks for real-time acquisition performance. +35% speed -50% latency
  • Configured Jetpack SDK and Linux-based scripting for embedded system automation, boot services, and operational monitoring.
  • Developed shell scripts for real-time system metrics monitoring, log analysis using grep/regex pipelines, and automated task management on field hardware.
Linux Administration & Shell Scripting
Personal Project · github.com/nilayjaltare7
Feb 2024 – Present Continuous Development

Ongoing hands-on Linux systems study — building production-quality shell utilities for system administration, monitoring, and log management.

  • Created shell scripts for real-time system metrics monitoring — CPU, memory, temperature, and process health dashboards.
  • Proficient in text processing and log analysis using grep, regex, awk, and sed pipelines for structured log extraction.
  • System management automation — cron-driven tasks, log rotation, disk usage monitoring, and service health checks.
  • Applied directly to production embedded Linux deployments on NVIDIA Jetson Nano and Raspberry Pi field hardware.
05

Technical Skills

Linux & Embedded Systems
Bash / Shell Scripting CPU Affinity (taskset) Freq Scaling systemd grep / awk / sed Log Analysis Process Mgmt Jetpack SDK L4T Raspberry Pi OS UART / AT Commands Docker on Linux
ML & Data Science
TensorFlow / Keras PyTorch Scikit-learn XGBoost LSTM Transformer Autoencoder (CAE) OpenCV Pandas / NumPy Matplotlib / Seaborn Anomaly Detection
AWS & Cloud IoT
AWS IoT Core Lambda SageMaker Bedrock EC2 / EBS S3 DynamoDB SNS / SQS Device Shadows EDA
Edge Hardware & IoT Protocols
NVIDIA Jetson Nano Raspberry Pi MQTT / TLS UART / Serial GPIO / I2C / SPI OTA Firmware Updates Real-Time Acquisition Sensor Telemetry
Programming & DevOps
Python C++ Shell Scripting React MSSQL Git / GitHub Docker Gunicorn
Desktop & Leadership
Tkinter / PyQt Executable Packaging Tech Lead Intern Mentorship Client Delivery Vendor Coordination Product Lifecycle
06

Projects

FEB 2025 – APR 2025
Remote Edge Device Management — Oil Industry
Bharat Flow Analytics · github.com/nilayjaltare7
  • Raspberry Pi edge data logger monitoring critical industrial parameters in field conditions.
  • Secure MQTT/TLS pipeline to AWS IoT Core — 99% transmission reliability in field deployments.
  • OTA firmware update system via AWS — 80% reduction in deployment time, no manual field visits.
  • AWS SNS-integrated alerting reduced incident response time by 40%.
  • Python automation scripts for edge-side acquisition, local anomaly detection, and DynamoDB/S3 batch logging.
  • Linux shell scripts for monitoring connectivity, managing services, and log rotation on Pi hardware.
Raspberry Pi Linux MQTT/TLS AWS IoT Core OTA Lambda DynamoDB SNS Python
OCT 2025 – JAN 2026
Claude Integration via Model Context Protocol (MCP)
Personal Project · LLM + Private Knowledge
  • Integrated Claude with Notion via MCP for context-aware querying of personal and technical knowledge bases — domain-specific retrieval without external web search.
  • Bridged LLM capabilities with structured private data sources, significantly improving accuracy of technical information retrieval.
  • Demonstrates understanding of AI system architecture, API integration patterns, and structured data pipelines.
Claude API MCP Notion LLM Integration Python REST API
DEC 2024 – MAR 2025
Hybrid LSTM-XGBoost Stock Price Prediction
github.com/nilayjaltare7
  • Modular prediction pipeline in Python processing 5+ years of historical data with TensorFlow/Keras and XGBoost.
  • LSTM for time-series feature extraction combined with XGBoost for robust regression — MAPE of 2.3%, R² of 0.94.
  • Significantly outperformed ARIMA and naive baseline models in robustness and prediction accuracy.
LSTM XGBoost TensorFlow Time-Series Python
MAY 2024 – JUL 2024
Image Colorization via Convolutional Autoencoders
github.com/nilayjaltare7
  • Skip-Loss CAE architecture achieving PSNR 28.7 dB and SSIM 0.89 — 15% improved colour accuracy over standard CAE.
  • Designed and benchmarked two deep learning architectures in PyTorch with convolutional layers and ReLU activations for spatial feature learning.
  • Demonstrates depth in generative deep learning and computer vision, applicable to NVIDIA Jetson vision pipeline contexts.
PyTorch Skip-Loss CAE Computer Vision CNN OpenCV
07

Certifications & Education

🏅
AWS Certified Machine Learning Engineer — Associate AWS Certificate Manager · Developing, deploying, and maintaining ML solutions on AWS 2025
🎓
Professional Certification in AI & Emerging Technologies IIT Hyderabad · AI, Blockchain, IoT & Quantum Computing — project-based 2024
📐
B.Tech — Electronics & Communication Engineering Minor in Artificial Intelligence & Machine Learning · MIT World Peace University, Pune 2025
🇩🇪
Goethe-Zertifikat B1 — German Language Goethe-Institut · Pune — demonstrates structured learning discipline and global readiness 2024
08

Contact

Open to the right opportunity.

I'm actively looking for roles in ML engineering, industrial IoT systems, and embedded Linux engineering — particularly with companies that build products where Linux runs close to the metal: NVIDIA, Qualcomm, NXP, Arm, or engineering-led startups in robotics, autonomous systems, or industrial AI. If you're building something that benefits from both edge hardware depth and cloud architecture experience, let's talk.

Get in touch
Phone +91 8600 400 581
Location Pune, India · Open to relocation
Open to
  • Embedded Linux / Systems Software Engineer roles
  • ML / Edge AI Engineer positions (NVIDIA, Qualcomm ecosystem)
  • Industrial IoT platform and architecture roles
  • Firmware / BSP Engineer — ARM/Jetson platforms
  • Robotics, autonomous systems, or industrial AI startups
  • Remote, hybrid, or Pune-based opportunities
  • Response within 24 hours