Skip to content
View jeypnet's full-sized avatar

Block or report jeypnet

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
jeypnet/README.md

Juan Pablo Pacheco: AI Solutions Engineer. AI agents, MCP and LLM integrations, workflow automation, and RF + AI software-defined radio with on-device machine learning

Claude MCP LLM APIs AI Agents Python PyTorch ONNX Runtime C# .NET Django Twilio Stripe Cloudflare GitHub Actions SDR / RF DSP

LinkedIn RF + AI projects

I turn customer and operational problems into AI integrations, automation, and working tools. My focus is AI solutions engineering: understand the workflow, design the solution, build with AI coding agents, test it, and own its operation. I also build RF + AI software: software-defined radio apps with on-device machine learning for drone detection, GPS interference, SCADA radio links, spectrum monitoring and signal classification.

At JoGo.Team, I build AI capabilities for a sports marketplace. My background in customer success, technical sales, and radio systems helps me connect business needs with implementation and explain the tradeoffs to the people using it.

Open to remote AI Solutions Engineer, AI Solutions Implementation, and AI Automation roles. Connect on LinkedIn.

Start here: problems I solve

Customer Account Intelligence
Customer success AI
Account health, churn risk, AI QBR briefs
Customer Win-Back
Lifecycle marketing automation
Engagement scoring, SMS and WhatsApp win-back
AI Design-to-Code Demo
AI sales engineering
Design-to-code demo for technical buyers
Area Project and evidence What it demonstrates
AI + sports marketplace JoGo MCP case study Connecting assistants to a real platform through structured tools
Customer operations + SaaS Account intelligence · sample report Risk prioritization, optional LLM briefs, and customer-facing reporting using synthetic accounts
Retention + lifecycle automation SMS / WhatsApp re-engagement Engagement scoring and channel selection, with a no-send demonstration
Radio + hardware integration SmartSDR client Network discovery, device commands, and asynchronous status callbacks
Radio + external data HF propagation checker Translating space-weather inputs into an explainable band-scoring heuristic
AI products + technical demonstrations Quest demos Evaluating Figma-to-React output and explaining integration tradeoffs to technical buyers

The repositories include demonstration code and technical utilities. Each README explains its scope and how to inspect it. Synthetic industry examples are not client references; employer experience does not imply deployment of these public projects at those employers or their customers.

✈ Flagship RF project: KG5ZZB SDR Airband Receiver

KG5ZZB SDR Airband Receiver: hear pilots, see the planes, read their messages, with an RTL-SDR dongle

A complete aviation radio app for a low-cost RTL-SDR dongle, built end to end in C# / .NET: an airband scanner with an AI noise filter, a live map of the planes around you from their ADS-B transponders, and the ACARS / VDL2 messages they send, explained in plain English. It runs offline, from your own antenna.

  • Own RTL-SDR driver (USB, tuner and PLL control in C#): the scanner checks every channel almost twice as often as with the free driver, with 1000 retunes in a row and no errors.
  • Own decoders: ADS-B, ACARS and VDL Mode 2 (D8PSK, Reed–Solomon), all in the app.
  • AI noise filter (RNNoise) for weak voices, and plug and play with two receivers: the app finds which one hears planes and gives it the map.
  • RTL-SDR, SDRplay and Airspy receivers · Windows, Mac and Linux.

⬇ Download free (Windows) · Mac & Linux · How it works and measured results

RF + AI: software-defined radio projects

Windows apps that turn a low-cost RTL-SDR into a sensor, with machine learning trained in PyTorch and run on-device with ONNX Runtime in C# / .NET. Each one is receive-only and ships with a model card that reports measured results against a classical baseline.

Drone RF Detector GNSS Interference Monitor
SCADA RF Link Monitor RF Signal Classifier
Spectrum Anomaly Detector IoT RF Inventory
Project Industry What it does Measured result
Drone RF Detector Counter-UAS, critical infrastructure Neural network classifies 900 MHz RF bursts; frequency-hopping and site-baseline alert logic; FAA Remote ID (ASTM F3411) over Bluetooth 96.4% accuracy vs 66.1% for rules
GNSS Interference Monitor Aviation, PNT, defense Maps GPS degradation from ADS-B NACp/NIC, estimates the jammer location, CNN classifies GPS L1 jammer types 99.2% vs 73.0% for rules
SCADA RF Link Monitor Energy, oil & gas, utilities (OT) Radio link health for RTU telemetry; autoencoder anomaly detection; MQTT and Modbus TCP outputs for SCADA/Ignition F1 0.82 vs 0.67 for thresholds
RF Signal Classifier RF machine learning, spectrum 1-D CNN names 14 modulations from raw IQ; Python/C# preprocessing parity; SigMF recording 90.2% at SNR ≥ 10 dB
Spectrum Anomaly Detector Spectrum monitoring, telecom Time-of-day baseline + autoencoder flag new, missing and interfering signals from 24 MHz to 1.7 GHz 90.3% recall at 0.1% false alarms
IoT RF Inventory IoT, physical/RF security Demodulates OOK/FSK sub-GHz devices, classifies modulation, DBSCAN fingerprinting, new-device alerts 97.6% vs 69.7% for rules

The RF + AI repositories are project showcases (description, results, model card); the source code is private and technical walkthroughs are available on request.

How I build

I use an AI-native development process. AI coding agents help produce the code; I own the problem definition, solution design, review, testing, troubleshooting, and completed workflow. Quest's repository explicitly identifies its generated platform output.

  • Discover and explain: connect customer requirements to a practical workflow, demonstrate the result, and discuss limitations.
  • Integrate: work with MCP, REST APIs, webhooks, messaging services, and operational data.
  • Operate and improve: investigate failures, evaluate fallback behavior, and use customer feedback and analytics to guide the next change.

Tools across my work include Claude Code, MCP, PyTorch, ONNX Runtime, C# / .NET (WPF), DSP and RTL-SDR, Anthropic and OpenAI APIs, Python/Django, Twilio, Stripe Connect, SendGrid, Cloudflare, GitHub Actions, Sentry, GA4, and Google Search Console. The projects above show specific slices of that work, not a claim that every tool is implemented in every repository.

Industry experience behind the projects

  • JoGo.Team: customer operations and AI integration for a sports marketplace.
  • BP3 Global: enterprise AI and automation sales across banking and manufacturing.
  • Konica Minolta: public-sector accounts, document workflows, and AI security platforms.
  • FlexRadio Systems: customer-facing work with software-defined radio across defense, utilities, and emergency response.
  • Quest: demonstrations of an AI design-to-code product for enterprise buyers.
  • Motorola BGH: hands-on radio and electronics support.

Engineering education at UTN FRBA, Extra Class amateur radio background, and bilingual English / Spanish communication inform how I approach technical problems.

LinkedIn · Browse the projects

Pinned Loading

  1. csm-account-intelligence csm-account-intelligence Public

    Customer success AI: explainable account health and churn-risk scoring, LLM-generated QBR briefs (OpenAI API), and customer-facing PDF health reports. Python workflow automation for SaaS customer o…

    Python

  2. flexradio-smartsdr-client flexradio-smartsdr-client Public

    Python client for the FlexRadio SmartSDR API: TCP command protocol, radio discovery and asynchronous status callbacks for software-defined radio hardware integration.

    Python

  3. hf-propagation-checker hf-propagation-checker Public

    HF radio propagation checker: pulls NOAA space-weather data (solar flux, Kp index) and scores HF band conditions with an explainable model. Python CLI with JSON output.

    Python

  4. Quest-Demos Quest-Demos Public

    AI design-to-code demo: Figma-to-React components and design-system themes generated with Quest.ai, with an evaluation of AI-generated frontend code for technical buyers.

    TypeScript

  5. twilio-whatsapp-reengagement twilio-whatsapp-reengagement Public

    Customer re-engagement automation: engagement scoring and smart channel selection across SMS and WhatsApp with the Twilio API. Python lifecycle-messaging workflow with a no-send CLI demo.

    Python