⚠️ Editor’s Note: This article was originally published by 563 on X (formerly Twitter). It is republished here with full credit to the author. All rights belong to the original author.
TL;DR – Each active subnet is matched to the kind of engineer or professional most likely to excel as a miner. Please note that my expertise does not fall into any of the below categories (I’m no ML engineer) and the following should just serve as an illustration of how vast the pool of knowledge is on these subnets.
LLM & LANGUAGE-MODELING
Goal: Continually improve the quality and capability of large language models through pre-training, fine-tuning, alignment and novel architectures.
- Apex (Subnet 1) – Language-model engineer, generation & reasoning | @MacrocosmosAI
- Templar (3) – Distributed-LLM ML engineer | @tplr_ai
- Pretrain (9) – Foundation-model pre-trainer | @MacrocosmosAI
- Dippy (11) – Model fine-tuner for companion bots | @dippy_ai
- Finetuning (37) – Alignment & continuous fine-tune specialist | @MacrocosmosAI
- Gradients (56) – AutoML engineer | @rayon_labs
- AI Factory (80) – Post-training engineer for niche models
CRYPTOGRAPHY & ZK
Goal: Produce provably correct ML inferences or transactions using zero-knowledge proofs and other cryptographic primitives.
- omron (2) – Zero-knowledge ML engineer | @omron_ai
PREDICTION & FORECASTING
Goal: Generate probabilistic predictions that are both well-calibrated and sharp for real-world events (sports, markets, climate, etc.).
- Infinite Games (6) – Probability modeler | @Playinfgames
- Zeus (18) – Climate & environmental data scientist | @zeussubnet
- Bettensor (30) – Sports probabilistic forecaster | @Bettensor
- SPORTSTENSOR (41) – Sports probabilistic forecaster | @sportstensor
- NextPlace (48) – Real-estate data scientist (CV knowledge would be helpful) | @NextPlace_AI
- Precog (55) – Crypto price-forecaster | @CM_Precog
- Gaia (57) – Earth-observation data scientist | @Gaia_AI_
- CheckerChain (87) – Product-review trust-score data scientist
TRADING & DEFI
Goal: Deliver risk-adjusted alpha by creating or executing quantitative strategies that outperform baseline benchmarks.
- Proprietary Trading Network (8) – Quantitative / prop trader | @taoshiio
- Sturdy (10) – DeFi yield strategist | @SturdyFinance
- Synth (50) – Volatility modeler | @modenetwork
- EfficientFrontier (53) – Quantitative / prop trader | @Effi_Frontier
- Alpha Trader Exchange (63) – Quantitative / prop trader (subnet token specific) | @_alphatraderx
COMPUTER VISION & MULTIMODAL
Goal: Produce accurate, real-time understanding of images, video and mixed-media streams for domain-specific tasks (sports, medical, deepfake).
- OMEGA Any-to-Any (21) – Vision-language / captioning engineer | @omegalabsai
- BitMind (34) – Deepfake-detection researcher | @BitMindAI
- Score (44) – Sports-video CV engineer | @webuildscore
- Safe Scan (76) – Medical-imaging CV engineer | @SAFESCAN_AI
- Vidaio (85) – AI-enabled video upscaler | @vidaio_
NLP – SPECIALIZED TASKS
Goal: Solve narrow, text-centric challenges such as embeddings, adversarial detection, segmentation or fact-checking at high precision and speed.
- ItsAI (32) – AI-text detection researcher | @ai_detection
- Chunking (40) – Token-optimization researcher | @chunking_subnet
- Condenses.ai (47) – Token-optimization researcher | @condensesai
- FakeNews (66) – Evidence-retrieval NLP engineer
- Vericore (70) – Evidence-retrieval NLP engineer
GENERATIVE MEDIA
Goal: Create high-fidelity synthetic content (3-D assets, images, websites) that meets validator quality tests for realism, consistency or artistic value.
- 404GEN (17) – text-to-3D modeling engineer | @404gen_
- Neural3D (46) – text-to-3D modeling engineer | @GoNeuralAI
- WebGenieAI (54) – text-to-application engineer
DATA COLLECTION & CURATION
Goal: Acquire, clean and label large, diverse, up-to-date datasets that applications, models, or other miners rely on.
- Data Universe (13) – Web-scraper / curator | @MacrocosmosAI
- OMEGA Labs (24) – Video-metadata scraper | @omegalabsai
- Real-Time Data by Masa (42) – Real-time data scraper | @getmasafi
- Dojo (52) – Crowdsourced labeling / data-generation | @TensorplexLabs
- ReadyAI (33) – Data structuring engineer | @ReadyAI_
- Patrol (81) – Crypto threat-intel / graph-ML engineer | @tao_dot_com
INFRASTRUCTURE & MLOPS / SRE
Goal: Keep compute, storage and nodes highly available, efficient and secure so the subnet can scale without downtime.
- SubVortex (7) – Subtensor node site-reliability engineer (SRE) | @SubVortexTao
- Celium (51) – GPU rental DevOps / SRE | @celiumcompute
- Compute Horde (12) – GPU rental DevOps / SRE | @ComputeHorde
- NI Compute (27) – GPU rental DevOps / SRE | @neural_internet
- Storb (26) – Storage SRE
- polariscloud.ai (49) – GPU marketplace operations | @polariscloudai
- Chutes (64) – MLOps platform engineer / SRE | @rayon_labs
- Hippius (75) – Decentralized storage & compute SRE | @hippius_subnet
ML SERVING & PERFORMANCE
Goal: Deliver model outputs at the lowest latency and cost through kernel-level optimization, batching and smart caching.
- Targon (4) – Low-latency LLM serving engineer | @manifoldlabs
- Nineteen.ai (19) – Full-stack inference engineer | @rayon_labs
- w.ai (39) – Edge device inference engineer | @WOMBO
AGENTS & AUTONOMOUS SYSTEMS
Goal: Build agent frameworks that can plan, act and learn to maximize task success in open or competitive environments.
- BitAgent – Rizzo (20) – Tool-calling agent engineer | @FrankRizz07
- web-agents (36) – RL web-automation engineer | @AutoppiaAI
- Agentao (62) – Code-generation agent engineer | @taoagents
- Agent Arena by Masa (59) – Social-agent developer (competitive) | @getmasafi
- Eastworld (94) – Agent-evaluation / robotics researcher | @Eastworld_AI
SECURITY & ADVERSARIAL ML
Goal: Identify and mitigate vulnerabilities, exploits or misinformation using ML-driven analysis and red-team techniques.
- Bitsec.ai (60) – Smart-contract security researcher | @bitsecai
- RedTeam (61) – Red-team code-challenge engineer | @_redteam_
- BitMind (34) – Deepfake / adversarial researcher (also in CV) | @BitMindAI
- ReinforcedAI Audits (92) – Solidity audit engineer (RL + AI) | @Reinforced_AI
BIOINFORMATICS & DRUG DISCOVERY
Goal: Accelerate the search for novel proteins, molecules and therapeutics via predictive modeling and large-scale simulation.
- Folding (25) – Protein-folding researcher | @MacrocosmosAI
- NOVA (68) – Drug-discovery scientist | @metanova_labs
GRAPH & MATHEMATICAL ML
Goal: Compute exact or approximate solutions to graph problems and symbolic math that traditional neural nets struggle with.
- LogicNet (35) – Mathematical-reasoning engineer
- Graphite (43) – Graph-algorithm engineer | @GraphiteSubnet
SEARCH & RETRIEVAL
Goal: Return relevant, trustworthy answers by ranking or retrieving documents with verifiable evidence.
- Desearch (22) – Semantic-search / retrieval engineer | @desearch_ai
MARKETING & GROWTH
Goal: Turn ML signals into measurable conversions to drive real sales and deliver value.
- BitAds (16) – Growth-marketing analyst | @BitAds_AI
- Bitcast (93) – Creators / educators | @Bitcast_network
Know someone that works in these fields and could be earning some sweet, sweet $TAO? Send them this list!
Be the first to comment