# Telnyx — Senior Machine Learning Engineer (Speech Synthesis)

| Field | Value |
|---|---|
| **Date found** | 2026-05-22 |
| **Company** | Telnyx |
| **Role** | Senior Machine Learning Engineer (Speech Synthesis) |
| **Location** | Dublin, County Dublin, Ireland — Fully Remote |
| **Salary** | Undisclosed |
| **Job URL** | https://www.linkedin.com/jobs/view/4315159934/ |
| **Status** | New |

---

## Company Research

| Field | Value |
|---|---|
| **Headquarters** | Chicago, Illinois, USA |
| **Founded** | 2009 |
| **Employees** | ~377 (379 on LinkedIn) |
| **LinkedIn** | https://www.linkedin.com/company/telnyx/ — 27,251 followers |
| **Website** | https://telnyx.com |
| **Blog** | — |

- **Product:** Full-stack AI infrastructure and global communications platform — private IP network, voice AI agents, edge compute, messaging, and APIs for real-time conversational AI.
- **Customers:** Businesses building real-time voice AI agents, global communications, and intelligent automation.
- **Notable:** Profitable and self-funded; owns private global multi-cloud IP network; building in-house speech synthesis for voice AI agents; 28% 2-year headcount growth.

---

## Job Summary

**What they do:** Telnyx builds the infrastructure layer for real-time AI — combining voice AI, global telephony, and edge compute into a single platform for conversational AI applications.

**The role:** Founding senior IC on the speech synthesis team, owning the full ML stack from training pipelines to low-latency inference for multilingual TTS models powering voice AI agents.

**Core work:**
- Design and implement training and inference pipelines for multilingual TTS (text-to-speech) models
- Build and fine-tune models using LLM-based, diffusion, and flow-matching architectures
- Engineer ultra-low-latency streaming speech generation (sub-100ms targets)
- Develop data pipelines for text, audio, and phonetic data at scale across dozens of languages

**Stack:** Python · PyTorch · Docker · Kubernetes · Ray · Kubeflow · MLflow · Kafka · Redis

**Work style:** Fully remote, Ireland-based. Founding member of TTS team — define the stack and practices from day one.

---

## Score: 62%

| Dimension | Score | Justification |
|---|---|---|
| Agentic AI depth (25%) | 10% | No agentic AI content whatsoever. This is speech synthesis / TTS engineering. LLMs are mentioned only as an architecture choice for TTS models, not for orchestration, agents, or autonomous pipelines. |
| Tech fit (25%) | 60% | Python, PyTorch fit well. Luca has audio ML background (ASR, multilingual, speaker recognition). Stack is familiar. However, the core work (TTS, diffusion models, speech synthesis) diverges from Luca's LLM/NLP/agentic specialisation. |
| Remote fit (25%) | 100% | Fully remote, Ireland-based. No on-site requirement. |
| Company culture fit (15%) | 70% | Small AI-infrastructure company (~377 employees), profitable, technically serious, building voice AI. Not AI-native in the agentic sense, but product-driven and engineering-led. |
| IC/leadership balance (10%) | 85% | Founding member of new team, full IC ownership of stack and practices from day one. No people management implied. |
| **Final (weighted)** | **62%** | Passes threshold driven by remote fit and IC ownership; agentic AI depth score is near-zero, which is the core weakness. |

---

## Strengths

- Fully remote Ireland — ideal location fit
- Founding team member role — full stack ownership and influence over architecture
- Profitable, self-funded company — no VC runway risk
- Luca has direct audio ML experience (multilingual ASR, speaker recognition, diarization)
- Voice AI agent context: TTS is infrastructure for the agentic voice layer Telnyx is building

---

## Weaknesses & Risks

- **Zero agentic AI content** — pure speech synthesis engineering, no LLM orchestration, no multi-agent systems
- Role goes in the opposite direction from Luca's target specialisation (agentic AI, not audio ML)
- Salary undisclosed; smaller company may offer below-target
- Speech synthesis is a very specialist domain — may lock Luca deeper into audio ML rather than agentic AI
- "Reposted 1 day ago" with 100+ applicants — significant competition already

---

## Suggestions

- Only pursue if agentic AI pipeline is slow — this is a safety net role given strong remote fit and IC ownership
- If applying, emphasise the multilingual ASR and audio ML background from the cybersecurity company
- Ask about the voice AI agent direction: is this TTS role a stepping stone toward agentic systems, or purely model training?
- Verify salary range before investing time in the process

---

## Interview Tracker

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