Local LLM-Based Agentic System for Social Media Content Generation and Analytics

In this project we designed and implemented a local-first, agentic social media content generation and analytics system. The platform includes a LangGraph orchestration layer, a locally hosted Qwen large language model, per-platform and per-business content strategies, a supervised multi-source curation graph, SQLite-based analytics computation, an Express REST API and a React dashboard, as well as authenticated posting to Twitter and LinkedIn. The system successfully demonstrates the feasibility of an end-to-end marketing workflow driven by open-weights inference, from ingesting raw datasets to data-grounded, model-generated insights. Future work includes completing the post-generation graph and its prompt engineering, expanding automated unit coverage to all graph nodes, collecting quantitative generation and insight-quality benchmarks, containerizing the deployment, fixing version-manager inconsistencies in the repo, and hardening the platform clients and authentication flows for production use.

[1] A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention Is All You Need,” Advances in Neural Information Processing Systems, vol. 30, 2017. doi: 10.48550/arXiv.1706.03762.    [2] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” in Proc. NAACL-HLT, 2019, pp. 4171–4186. doi: 10.18653/v1/N19-1423.    [3] T. B. Brown et al., “Language Models are Few-Shot Learners,” Advances in Neural Information Processing Systems, vol. 33, 2020. doi: 10.48550/arXiv.2005.14165.    [4] E. J. Hu, Y. Shen, P. Wallis, Z. Allen-Zhu, Y. Li, S. Wang, L. Wang, and W. Chen, “LoRA: Low-Rank Adaptation of Large Language Models,” 2021. doi: 10.48550/arXiv.2106.09685.    [5] T. Dettmers, A. Pagnoni, A. Holtzman, and L. Zettlemoyer, “QLoRA: Efficient Finetuning of Quantized LLMs,” 2023. doi: 10.48550/arXiv.2305.14314.    [6] L. Ouyang et al., “Training Language Models to Follow Instructions with Human Feedback,” Advances in Neural Information Processing Systems, vol. 35, 2022, pp. 27730–27744. doi: 10.52202/068431-2011.    [7] Y. Wang, Y. Kordi, S. Mishra, A. Liu, N. A. Smith, D. Khashabi, and H. Hajishirzi, “Self-Instruct: Aligning Language Models with Self-Generated Instructions,” in Proc. 61st Annual Meeting of the Association for Computational Linguistics, 2023, pp. 13484–13508. doi: 10.18653/v1/2023.acl-long.754.    [8] R. Rafailov, A. Sharma, E. Mitchell, C. D. Manning, S. Ermon, and C. Finn, “Direct Preference Optimization: Your Language Model is Secretly a Reward Model,” Advances in Neural Information Processing Systems, vol. 36, 2023. doi: 10.52202/075280-2338.    [9] A. Yang et al., “Qwen2 Technical Report,” 2024. doi: 10.48550/arXiv.2407.10671.    [10] A. Yang et al., “Qwen2.5 Technical Report,” 2024. doi: 10.48550/arXiv.2412.15115.