Jakarta & Bandung, Indonesia • Open to Remote AI & Engineering Work

Alfi Muhammad Akbar Baria

AI Trainer & LLM Evaluator • Embedded Systems Engineer • Smart Metering Lead

Hi, I'm Alfi. For over ten years, I've designed custom PCBs, microcontroller firmware, and industrial metering solutions for teams in automotive, energy, and public utilities. Since 2021, I've also worked with global AI teams to test large language models, rank competing responses side by side, and catch factual or cultural errors in Indonesian and English.

Building physical hardware teaches you to respect small details. When a board trace is noisy or a sensor reading drifts, something real stops working in the field. I bring that same practical habit to AI quality work: reading user prompts carefully, checking claims against real sources, and writing clear notes on why one answer works better than another.

See Hardware & AI Projects ↓ Contact Details LinkedIn Profile ↗

Live Bench & QA Monitor

60A PWM / UART

CH1: Gold 60A PWM Duty 72% CH2: Silver MAX6675 Thermal

Active Focus KiCad / EAGLE + Embedded C++

Field Interfaces ESP32 Wi-Fi AP • SPI / I2C / UART

Verified Output Pertamina SmartMT & Alphard RSE

10+ yrs

PCB design, embedded C++, IoT hardware, and technical sales

10,000+

AI annotation & search evaluation tasks across 100+ task types

>4.75/5

Average QA score on Indonesian LLM evaluations (2026)

+20%

Sales growth contribution in smart kWh & water metering

KiCad & CadSoft EAGLE PCB RLHF & Side-by-Side LLM Ranking Embedded C++ • ESP32 / ESP8266 Meridial Marketplace by Invisible (>4.75/5 QA) 60A BTS7970 PWM Motor Drivers Multimodal Image & Localization QA Smart kWh & Prepaid Water Metering SPI • I2C • UART Industrial Telemetry KiCad & CadSoft EAGLE PCB RLHF & Side-by-Side LLM Ranking Embedded C++ • ESP32 / ESP8266 Meridial Marketplace by Invisible (>4.75/5 QA) 60A BTS7970 PWM Motor Drivers Multimodal Image & Localization QA Smart kWh & Prepaid Water Metering SPI • I2C • UART Industrial Telemetry

Core Capabilities

Three areas where I do hands-on work

My day-to-day background combines circuit design on the bench, field coordination with utility customers, and structured evaluation of language and image models.

01 • AI QUALITY

LLM & Multimodal AI Evaluation

I test how models handle real user requests, multi-turn memory, grounded facts, and Indonesian local nuance. Instead of vague feedback, I write concrete, turn-by-turn rationales explaining what went wrong and how to fix it.

RLHF & SxS Ranking Hallucination Checks AI Personalization Multimodal Image QA Indonesian Localization

02 • HARDWARE

Embedded Systems & PCB Design

From schematic capture to SMD board assembly and C++ firmware, I build reliable controllers that read industrial sensors, drive high-current motors and pneumatic valves, and talk to Android tablets or local Wi-Fi web servers.

KiCad & EAGLE Embedded C++ ESP32 / ESP8266 SPI / I2C / UART 60A PWM Drivers

03 • UTILITY SOLUTIONS

Technical Sales & Smart Metering

I work directly between utility clients like PLN, government certification bodies, and internal R&D engineers. That includes preparing tender proposals, running field training, and resolving technical issues on smart kWh and water meters.

Smart kWh Meters Prepaid Water Meters Utility Tenders After-Sales Training Regulatory Docs

AI Quality & RLHF

How I evaluate language and multimodal models

Good AI evaluation requires more than picking which paragraph looks nicer. Models often write confident answers that miss a time constraint, invent a detail, or sound unnatural to a native Indonesian reader.

Since 2021, I have worked across OneForma/UHRS, Outlier, and Meridial Marketplace (by Invisible) on evaluation workflows that test factual grounding, multi-turn personalization, and visual prompt alignment.

Languages Rated

Bahasa Indonesia Native / Bilingual
English Full Professional
Sundanese Native / Regional Nuance

Meridial Marketplace, by Invisible

Multilingual Indonesian LLM Quality

Review Indonesian model outputs for local user intent, tone, response length, and automated-judge accuracy across four workstreams, maintaining an average QA score above 4.75 out of 5.00.

Multi-Turn Benchmarking

AI Personalization & Memory

Tested long multi-turn conversations for user preference retention and contextual reasoning, logging exact turn-level evidence whenever a model dropped prior context or misread dates.

Outlier Projects

RLHF Ranking & Targeted Rewrites

Designed challenging test prompts, verified source groundedness and instruction compliance, ranked competing model outputs side by side, and rewrote flawed responses.

Visual & Search Quality

Multimodal Image & Search Rating

Assessed hundreds of AI-generated images against dual reference photos for visual artifacts and prompt fidelity (QA >4.5/5), building on 10,000+ search relevance tasks at OneForma/UHRS.

Hardware & Field Footage

Project video demonstrations

Short bench and in-vehicle clips showing physical boards, motor drivers, touchscreen interfaces, and wireless control workflows in operation.

3 Embedded MP4 Demonstrations

SmartMT Wireless Compartment Control

Demonstration of smartphone browser control over local Wi-Fi AP triggering pneumatic solenoid valves on fuel tanker compartments.

Cabin TV Lift, PDLC Glass & Seat Automation

In-vehicle test of the custom RSE main board and 60A PWM motor driver operating motorized partitions, VIP privacy glass, and ambient lighting.

Touchscreen Serial Link & PCB Bench Bring-Up

Hardware bench verification of UART touchscreen commands switching relay banks, smart film doors, and power inverters.

Direct Contact

Let's talk about your project or team

Whether you need an Indonesian/English AI quality evaluator, custom microcontroller and PCB development, or technical leadership on smart metering solutions, feel free to reach out directly.

[email protected]

Send Email

Phone & WhatsApp

+62 813-1307-5521

linkedin.com/in/alfibaria89

Open Profile