Skip to content
Bima Marta Sugiharto
Available for software engineering roles Bandung, Indonesia (UTC+7 / WIB)

Software
engineer
building systems.

Software engineer building production web applications, distributed queue architectures, high-performance automation CLI tools, and computer vision pipelines. Experienced in TypeScript, Python, Node.js, BullMQ, Valkey, and deep learning systems.

Distributed queuesCLI automationComputer vision
Download CV Explore projects
GitHub ↗ LinkedIn ↗ WhatsApp · 08999102449 ↗
Bima Marta Sugiharto

Bima Marta Sugiharto

Bandung, Indonesia (UTC+7 / WIB)

Available

Selected
work.

Products, tools and research shaped by software engineering, automation and computer vision.

Taktika AI
Commercial / Ideact Live

Taktika AI

AI football intelligence platform combining video ingestion, 2D radar, tactical heatmaps, shot maps, and coach assistant chat.

Next.jsTypeScriptComputer VisionTailwind CSSValkeyPython
Input
Match video
Output
2D radar, heatmaps, shot maps
  • Ingests match recordings to generate tactical top-down 2D field radars and player trajectories.
  • Processes player tracking data into spatial heatmaps, pass network graphs, and shot probability maps.
  • Includes an embedded conversational coach assistant for contextual match query analysis.
Open project ↗
InHire AI
Commercial / Ideact Live

InHire AI

Screening platform for high-volume recruitment running automated CV evaluations, candidate scoring, and video interviews.

TypeScriptNode.jsNext.jsBullMQValkeyPostgreSQLDrizzle ORM
Use case
High-volume screening
Queue
BullMQ + Valkey
  • Processes candidate resumes through structured parsing models to rank qualified profiles.
  • Maintains transactional safety via atomic balance and credit consumption ledgers.
  • Automates interview dispatch, candidate calendar syncing, and background processing via BullMQ.
Open project ↗
bima / txbot

$ python solana_buy_bot.py -kf keys.txt --auto-find -mc 4000 -tl 2 -f low -bar 0.09-0.1

🚀 Solana Buy Bot Starting...

Number of wallets: 2 | Number of loops: 1 | Buy amount range: 0.09-0.1 SOL

Found 6 tokens with market cap < $4000

Buying token: TOKEN (mint...) | SOL Amount: 0.094 | Fee Level: low

Getting quote...

Input: 0.094 SOL | Output: 18,402 tokens | Price Impact: 0.4100%

Using low priority fee level, sending transaction...

✅ Transaction sent, explorer link: solscan.io/tx/...

Selling token: keeping a random 50-100 tokens, sending sell transaction...

Closed Source Private project

TXBot

Private Solana trading CLI. It finds low market cap tokens from the Raydium API, buys them across several wallets, then sells most of the position after a delay.

PythonSolanasoldersRaydium APIClickCLI
  • Private repository. The terminal replay shows the real log format with sample values.
  • Takes wallet keys from a flag or a keys file and runs a configurable number of loops.
  • Supports priority fee levels, slippage in basis points, buy amount ranges, and buy and sell delays.
Private repository
bima / ticket-com-automation

$ python run.py

🎫 Tiket.com Auto Order Bot

? Pilih mode: Mode 2: Auto pembelian by slug

✅ [1] Valid session: account loaded from cookiestiket.txt

📦 Auto-selected: 'Category 1' (keyword: cat 1)

⏳ [ 09:59:58 ] Countdown: 2 detik

🚀 Waktu tercapai

✅ Captcha solved

✅ Order dibuat: <order_id>

💳 Payment Info: URL sent to Telegram

Closed Source Private project

ticket-com-automation

Private ticket checkout CLI for tiket.com events. It holds sessions for several accounts, waits for the sale to open on a countdown, solves the captcha, places the order, and sends the payment link to Telegram.

Pythoncurl_cffiRichquestionaryTelegram Bot APICLI
  • Private repository. The terminal replay shows the real log format with sample values.
  • Six modes: login, auto checkout by slug, ticket URL extraction, extraction plus checkout, quota check, and manual cookies.
  • Uses curl_cffi browser impersonation, Rich console output, and questionary prompts.
Private repository
Football Movement Analysis System
Research / Undergraduate Thesis Completed

Football Movement Analysis System

Computer vision pipeline integrating YOLO 26 for player detection, YOLO 26 Pose for field keypoints, homography transformation for pitch coordinate mapping, ByteTrack for tracking, and SigLIP2 + UMAP + GMM for team classification.

PythonPyTorchYOLO 26ByteTrackSigLIP2UMAPOpenCV
Player mAP@50
97.2%
Keypoint mAP@50
96.9%
Team Silhouette
86.81%
  • Detection and tracking: Custom YOLO 26 architecture achieved 97.2% mAP@50 for player detection and 96.9% mAP@50 on 32 field keypoints.
  • Homography projection: Mapped camera perspective coordinates onto standard 105x68 meter pitch coordinates.
  • Team classification: Extracted visual embeddings with SigLIP2, reduced dimensionality via UMAP, and partitioned teams using Gaussian Mixture Models achieving 86.81% silhouette score.
  • Visual analytics: Generated positional heatmaps, dynamic tactical lines, and trajectory history.
Explore research ↓
USAGII
Closed Source Live

USAGII

Unified AI API gateway and Telegram billing integration providing pay-as-you-go access to language models with streaming edge completions.

TypeScriptCloudflare WorkersTelegram Bot APIEdge RuntimeTailwind CSS
Runtime
Cloudflare Workers
Billing
Pay-as-you-go credits
  • Routes requests through Cloudflare Workers at the edge.
  • Automated balance top-ups, invoice delivery, and command-driven usage monitoring over Telegram Bot API.
  • Provides high-availability failover across multiple model providers.
Open project ↗
3craft.studio
Closed Source In Progress

3craft.studio

Creative 3D studio platform and custom asset modeling pipeline delivering precision print-ready models.

AstroTypeScriptThree.jsTailwind CSS
  • Interactive 3D model showcase with client asset submission interface.
  • Print parameter estimation and digital asset catalog management.
Open project ↗
Mobx
Closed Source In Progress

Mobx

Market-discovery and token launch-draft terminal for web3 assets built on Next.js, Tailwind, and EVM contract integrations.

Next.jsTypeScriptEVM / Web3Tailwind CSSViem
  • Live decentralized exchange orderbook scanning and liquidity health meters.
  • Contract verification inspection and interactive token launch dashboard.
Open project ↗
sofascore-py
Open Source Open Source

sofascore-py

Python package to fetch, parse, and structure live match statistics, player ratings, and shot events into analysis-ready formats.

PythonWeb ScrapingPandasData Analytics
  • Structured extraction of raw match event streams and player heatmaps.
  • Clean tabular pandas DataFrame exports for downstream research and analytics.
Source on GitHub ↗
Convert-It
Open Source Open Source

Convert-It

Cross-platform desktop media converter wrapping FFmpeg using Tauri and TypeScript for local file transformations.

TauriRustTypeScriptFFmpegReact
  • Hardware-accelerated local video, audio, and image format conversion.
  • Zero cloud transmission ensuring complete user file privacy.
Source on GitHub ↗
WebSeek
Open Source Open Source

WebSeek

Fast web information retrieval and exploration engine built for query indexing and rapid result synthesis.

TypeScriptNext.jsVercelSearch API
Open project ↗

Where I’ve
worked.

Engineering roles across production software, public services and industrial systems.

Ideact mark

Ideact

PT Akal Cipta Teknologi

Software Engineer

Jan 2026 to Present · Bandung, Indonesia (Hybrid)

Company website ↗

Building core platforms for InHire AI and Taktika AI products. Designing asynchronous queue processing pipelines, atomic balance ledgers, and video intelligence interfaces.

  • Developed core platform features for InHire AI including atomic credit transaction ledgers, candidate assessment workflows, and candidate progress tracking.
  • Designed asynchronous queue workers using BullMQ and Valkey for email dispatch, webhook triggers, and external calendar event synchronizations.
  • Constructed calendar scheduling workflows with automated iCalendar payload generation for direct email interview invitations.
  • Built pitch analysis visualization interfaces for Sportsmeister, integrating client-side video preview validation with frame extraction.
TypeScriptNode.jsNext.jsBullMQValkey / RedisPostgreSQLDrizzle ORMTailwind CSS
Dinas Sosial Kota Bandung mark

Dinas Sosial Kota Bandung

Fullstack Web Developer Intern

Mar 2025 to Apr 2025 · Bandung, Indonesia

Company website ↗

Constructed internal administrative tools, automated content collection, and public portal caching layers for municipal social service operations.

  • Engineered an administrative content management system enabling staff to manage dynamic public announcements, service guidelines, and verified data entries.
  • Built an automated Instagram data extraction script in Python to gather departmental activity media and sync feeds to the public portal.
  • Architected multi-tier caching layers and optimized relational database queries to maintain portal availability during traffic spikes.
PHPLaravelPythonMySQLJavaScriptTailwind CSS
PT Len Industri (Persero) mark

PT Len Industri (Persero)

Fullstack Web Developer Intern

Mar 2022 to May 2022 · Bandung, Indonesia

Company website ↗

Developed RESTful data endpoints, operational dashboard integrations, and automated testing contracts for state enterprise industrial systems.

  • Implemented secure RESTful API endpoints for internal industrial data management services.
  • Connected third-party data providers and internal system feeds into unified monitoring dashboards for operational teams.
  • Authored automated unit and integration tests to enforce data contract integrity and prevent regression across schema updates.
JavaScriptNode.jsExpressPostgreSQLJestBootstrap

Research / Undergraduate thesis

Reading the game through computer vision.

Watch full demo ↗

Player tracking / Pitch homography / Tactical movement

Computer vision pipeline integrating YOLO 26 for player detection, YOLO 26 Pose for field keypoints, homography transformation for pitch coordinate mapping, ByteTrack for tracking, and SigLIP2 + UMAP + GMM for team classification.

  • Detection and tracking: Custom YOLO 26 architecture achieved 97.2% mAP@50 for player detection and 96.9% mAP@50 on 32 field keypoints.
  • Homography projection: Mapped camera perspective coordinates onto standard 105x68 meter pitch coordinates.
  • Team classification: Extracted visual embeddings with SigLIP2, reduced dimensionality via UMAP, and partitioned teams using Gaussian Mixture Models achieving 86.81% silhouette score.
  • Visual analytics: Generated positional heatmaps, dynamic tactical lines, and trajectory history.
Player mAP@50
97.2%
Keypoint mAP@50
96.9%
Team Silhouette
86.81%

Credentials.

University records, technical assessments and professional development.

Web Development

WIB-U Volume 2: Crafting Memorable Design Thinking

HIMASI & CREAVID UNIBI

View credential (png) ↗
Backend & Cloud

SQL and Relational Databases 101

IBM Cognitive Class

Credential ID: b669457a764b4008ac615894db7949c5

View credential (pdf) ↗
Cybersecurity

Cybersecurity and Data Protection

FTI Universitas Informatika dan Bisnis Indonesia

Credential ID: 017/WS.CS/FTI.UNIBI/VI/2025

View credential (pdf) ↗
Backend & Cloud

Blockchain Masterclass

Centre for Finance, Technology and Entrepreneurship (CFTE)

Credential ID: f73dbphfuu

View credential (pdf) ↗
AI & Data Science

IDCamp x Dicoding Live #9: Unlock Limitless AI with Agentic Trend

Dicoding Indonesia & Indosat Ooredoo Hutchison

View credential (pdf) ↗
AI & Data Science

Building GenAI Apps Learning Badge Path

MongoDB University

Credential ID: MDB2c0yrtbwvp

View credential (pdf) ↗
AI & Data Science

Using Vector Search for Semantic Search

MongoDB University

Credential ID: MDBxjaactkafq

View credential (pdf) ↗
AI & Data Science

Using Atlas Vector Search for RAG Applications

MongoDB University

Credential ID: MDB3xw2puzfe5

View credential (pdf) ↗

Talk / Juli 2024

Public speaking
& workshops.

Pemateri (Speaker) · Bandung, Indonesia (Live via Zoom)

Hosted by Program Studi Sistem Informasi, Universitas Informatika dan Bisnis Indonesia (UNIBI)

Belajar Artificial Intelligence Lewat Sepak Bola

Bagaimana Komputer Bisa Menonton Pertandingan?

Webinar interaktif membahas implementasi Computer Vision dan Artificial Intelligence dalam analisis pertandingan sepak bola: pelacakan pemain dan bola (object detection & multi-object tracking), deteksi kejadian di lapangan, serta pemetaan koordinat pitch.

  • Konsep dasar Computer Vision dan AI dalam video pertandingan olahraga
  • Deteksi serta pelacakan real-time pemain, wasit, dan pergerakan bola
  • Ekstraksi statistik pergerakan dan proyeksi koordinat lapangan sepak bola
Original announcement ↗
Official UNIBI announcement: Belajar Artificial Intelligence Lewat Sepak Bola