# Tether Operations Limited — AI Research Engineer (Agentic Post-training)

| Field | Value |
|---|---|
| **Date found** | 2026-06-01 |
| **Company** | Tether Operations Limited |
| **Role** | AI Research Engineer (Agentic Post-training) |
| **Location** | Remote (100% Worldwide) |
| **Salary** | Undisclosed |
| **Job URL** | https://to.indeed.com/aagbhsdkkrw4 |
| **Status** | New |

---

## Company Research

| Field | Value |
|---|---|
| **Headquarters** | Road Town, British Virgin Islands (operations globally distributed) |
| **Founded** | 2014 |
| **Employees** | ~100–500 (lean distributed team globally) |
| **LinkedIn** | linkedin.com/company/tether-to |
| **Website** | tether.to |
| **Blog** | tether.to/en/news |

- **Product:** Tether is the issuer of USDT (world's largest stablecoin by market cap); "Tether Data" is their AI division building edge-device AI models, KEET (private P2P communication), and AI infrastructure for fintech/Web3.
- **Customers:** Crypto exchanges, wallets, payment processors; hundreds of millions of USDT users worldwide.
- **Notable:** USDT has ~$130B+ circulating supply; Tether is one of the most profitable companies per employee globally; Tether Data is a fast-growing AI R&D arm focused on on-device frontier models.

---

## Job Summary

**What they do:** Tether's AI division builds frontier LLMs with agentic capabilities optimised for edge devices (smartphones), advancing post-training methodologies for tool use and autonomous reasoning.

**The role:** Senior IC Research Engineer driving post-training of agentic and tool-using LLMs to achieve SOTA results, with end-to-end ownership from data curation to evaluation and model release.

**Core work:**
- End-to-end post-training research: tool use fine-tuning, function calling, RLHF/RLAIF on multi-turn agentic interactions
- Curate agentic training data (tool-use trajectories, reasoning chains, environment interactions) and build evaluation suites
- Drive model improvements in multi-agent coordination, reasoning calibration, and factuality at scale

**Stack:** Python · PyTorch · LLMs (post-training at scale) · Distributed training (multi-node GPU) · Evaluation frameworks · Hugging Face · RLHF/DPO · Multi-agent coordination

**Work style:** 100% remote worldwide, lean team, fast-moving crypto/AI company; strong publication record expected.

---

## Score: 77%

| Dimension | Score | Justification |
|---|---|---|
| Agentic AI depth (25%) | 90% | Core mission is post-training for agentic tool use, planning, and autonomous multi-step tasks — the deepest possible agentic AI research scope |
| Tech fit (25%) | 50% | Python/PyTorch/LLMs overlap well; but heavy research track (distributed training, RL post-training) and publication requirement are gaps; no LangChain/LangGraph use |
| Remote fit (25%) | 100% | 100% remote worldwide — perfect |
| Company culture fit (15%) | 55% | Fast-moving, lean, autonomous — fits profile; crypto domain is niche; Tether Data AI division has a genuine research mission |
| IC/leadership balance (10%) | 85% | Pure IC research engineer role, no management, high technical autonomy |
| **Final (weighted)** | **77%** | |

---

## Strengths

- Deepest agentic AI scope in the index: post-training for tool use and autonomous reasoning
- 100% remote worldwide — no location constraints at all
- Lean, high-impact team with significant technical autonomy
- Tether Data is building genuine frontier AI, not wrappers
- USDT's financial resources mean strong R&D investment capacity

---

## Weaknesses & Risks

- ⚠️ Potential duplicate of #009 (Rejected) — very similar role title and scope (AI Research Engineer, Agentic Post-training); #009 was posted via Jobgether for undisclosed client
- ⚠️ Salary undisclosed — Tether is opaque on compensation
- Requires MS/PhD preferred + publications at top-tier conferences (NeurIPS, ICML, ICLR etc.) — significant gap vs. Luca's profile
- Pure research track: distributed training at scale, RL post-training — further from production engineering strength
- Crypto/Web3 domain may not align with career trajectory preference
- Tether has faced regulatory scrutiny regarding USDT reserves (reputational risk)

---

## Suggestions

- Only apply if comfortable with the research track and publication expectations
- Emphasise production agentic pipeline work at Fenergo as applied post-training relevance
- Ask about the distinction between research and engineering roles — there may be a more applied position
- Verify role is not a repost of #009 before spending time on application

---

## Interview Tracker

| Stage | Date | Notes |
|---|---|---|
| Applied | | |
| Recruiter screen | | |
| Technical interview | | |
| Final round | | |
| Offer / Outcome | | |