Felipe 0liveira

AI Engineer with a background spanning some of Brazil's biggest fintechs and tech companies — Nubank, PagSeguro, Serasa, and Gympass (Wellhub) — now Staff AI Engineer at Telus Digital, where I design

Brazil Joined on Aug 18, 2019 https://felipeoliveira.xyz

Work

Staff AI Engineer at TELUS Digital

Felipe 0liveira

Oct 16

Distillation's Hidden Costs and Payoffs — Teachers Leak Behavior, Reward Models Surpass Them

#ai #llm #machinelearning

5 min read

Want to connect with Felipe 0liveira?

Create an account to connect with Felipe 0liveira. You can also sign in below to proceed if you already have an account.

Create Account

Already have an account? Sign in

Felipe 0liveira

Oct 15

Perfectly Consistent, Still Wrong — LLM Judges Keep Failing the Same Way, Plus Cheaper Alternatives and Unified Observability

#ai #llm #machinelearning #monitoring

5 min read

Felipe 0liveira

Oct 14

Multi-Turn Memory, Mixed Precision, and Safer Rollouts — Serving Catches Up to Agentic Workloads

#agents #ai #llm #machinelearning

8 min read

Felipe 0liveira

Oct 13

Agents Are Writing Their Own Playbooks — and Occasionally Running Off With Them

#agents #ai #claude #llm

8 min read

Felipe 0liveira

Oct 12

Agent Memory Explodes While RAG's Favorite Assumptions Get a Reality Check

#agents #ai #llm #rag

10 min read

Felipe 0liveira

Oct 11

Stacking Guardrails Won't Save You — This Week's AI Security Research Proves It Again

6 min read

Felipe 0liveira

Oct 10

Haiku gets cheaper (but hungrier) as Mistral swings back toward the frontier

#news #ai #llm #machinelearning

5 min read

Felipe 0liveira

Oct 9

GRPO leaves the chatbot and starts fixing OCR, while DPO and GRPO get debugged from the inside

#ai #deeplearning #llm #machinelearning

6 min read

Felipe 0liveira

Oct 8

Judges Learn to Compress, Teams Learn to Trace — This Week in Evals

#ai #llm #machinelearning #nlp

5 min read

Felipe 0liveira

Oct 7

The KV Cache Is the Bottleneck Now — 1-Bit Quantization, MoE Stragglers, and Per-Second GPU Billing

#ai #cloud #llm #performance

5 min read

Felipe 0liveira

Oct 6

The Agent Harness Goes Platform-Native — and the Guardrails Scramble to Keep Up

#agents #ai #api #openai

9 min read

Felipe 0liveira

Oct 5

Cheaper Graphs, Bigger Risks: GraphRAG's Efficiency Push Meets Its First Security Scare

#ai #llm #rag #security

9 min read

Felipe 0liveira

Oct 4

When Agents Overreach — This Week's LLM Security News Is All About Tool-Calling Gone Wrong

5 min read

Felipe 0liveira

Aug 25

Passing Once Isn't Reliable — This Week's Agent Engineering Puts the Harness Before the Model

#agents #ai #llm #softwareengineering

6 min read

Felipe 0liveira

Aug 24

Ingest-Time Compilation Takes On Query-Time RAG, and Agentic Retrieval Meets Its Limits

#ai #llm #machinelearning #rag

6 min read

Felipe 0liveira

Aug 9 '25

🚀 SemVer: por que esse `1.2.3` muda sua vida (e seus bugs)

#semver #semanticversion #webdev #programming

1 min read

Felipe 0liveira

Mar 27 '25

Definitivamente pontuando tasks

#storypoint #agile #scrum #kanban

3 min read

Felipe 0liveira

Nov 18 '24

Iniciando na área de desenvolvimento (guia prático)

#webdev #javascript #beginners #programming

6 min read

Felipe 0liveira

Aug 17 '20

Liberar espaço deletando os node_modules/

#npm #node #modules #linux

1 min read

Felipe 0liveira

Apr 14 '20

Frontend - Testes E2E

#e2e #protractor #javascript #jasmine

1 min read

Felipe 0liveira

Apr 14 '20

Sua alma na documentação - DocSoul

#documentacao #javascript #docsoul #docummentation

2 min read

Felipe 0liveira

Mar 25 '20

[MacOS] Biometria para 'sudo' no terminal

#macos #sudo #fingerprint #terminal

2 min read

Felipe 0liveira

Nov 5 '19

Visual Studio Online - Minhas considerações

#vscode #vsonline #developmnt

1 min read

Felipe 0liveira

Aug 26 '19

CSS Buttons - Dopamina

#codepen #css #buttons #design

1 min read

DEV Community

Felipe 0liveira

1 Week Community Wellness Streak

Seven Year Club

Six Year Club

Five Year Club

Writing Debut

Four Year Club

Three Year Club

Two Year Club

One Year Club

GitHub Repositories

Skills/Languages

Currently learning

Currently hacking on

Distillation's Hidden Costs and Payoffs — Teachers Leak Behavior, Reward Models Surpass Them

Want to connect with Felipe 0liveira?

Perfectly Consistent, Still Wrong — LLM Judges Keep Failing the Same Way, Plus Cheaper Alternatives and Unified Observability

Multi-Turn Memory, Mixed Precision, and Safer Rollouts — Serving Catches Up to Agentic Workloads

Agents Are Writing Their Own Playbooks — and Occasionally Running Off With Them

Agent Memory Explodes While RAG's Favorite Assumptions Get a Reality Check

Stacking Guardrails Won't Save You — This Week's AI Security Research Proves It Again

Haiku gets cheaper (but hungrier) as Mistral swings back toward the frontier

GRPO leaves the chatbot and starts fixing OCR, while DPO and GRPO get debugged from the inside

Judges Learn to Compress, Teams Learn to Trace — This Week in Evals

The KV Cache Is the Bottleneck Now — 1-Bit Quantization, MoE Stragglers, and Per-Second GPU Billing

The Agent Harness Goes Platform-Native — and the Guardrails Scramble to Keep Up

Cheaper Graphs, Bigger Risks: GraphRAG's Efficiency Push Meets Its First Security Scare

When Agents Overreach — This Week's LLM Security News Is All About Tool-Calling Gone Wrong

Passing Once Isn't Reliable — This Week's Agent Engineering Puts the Harness Before the Model

Ingest-Time Compilation Takes On Query-Time RAG, and Agentic Retrieval Meets Its Limits

🚀 SemVer: por que esse `1.2.3` muda sua vida (e seus bugs)

Definitivamente pontuando tasks

Iniciando na área de desenvolvimento (guia prático)

Liberar espaço deletando os node_modules/

Frontend - Testes E2E

Sua alma na documentação - DocSoul

[MacOS] Biometria para 'sudo' no terminal

Visual Studio Online - Minhas considerações

CSS Buttons - Dopamina